A grammar of Pite Saami

Pite Saami is a highly endangered Western Saami language in the Uralic language family currently spoken by a few individuals in Swedish Lapland. This grammar is the first extensive book-length treatment of a Saami language written in English. While focussing on the morphophonology of the main word classes nouns, adjectives and verbs, it also deals with other linguistic structures such as prosody, phonology, phrase types and clauses. Furthermore, it provides an introduction to the language and its speakers, and an outline of a preliminary Pite Saami orthography. An extensive annotated spoken-language corpus collected over the course of five years forms the empirical foundation for this description, and each example includes a specific reference to the corpus in order to facilitate verification of claims made on the data. Descriptions are presented for a general linguistics audience and without attempting to support a specific theoretical approach, but this book should be equally useful for scholars of Uralic linguistics, typologists, and even learners of Pite Saami.


Introduction
This description of the grammar of the Pite Saami language is intended to provide a general linguistics audience with an overview of phonological and morphosyntactic structures found in the language. It is based on a corpus of spoken language data collected over several years, including both free speech and elicited data. It covers the phonology, morphology and phrasal structures, and how these interact with each other, while also providing a sketch of the clausal phenomena.
The present chapter provides background information on Pite Saami. It includes an overview of the language and its speakers in §1.1, then provides details about the documentation project on which this study is based in §1.2, and finally presents a typological profile in §1.3.

Linguistic genealogy
The Saami languages form a sub-branch of the Uralic language family. Pite Saami is classified as the southern-most Western Saami language in the Northern group. Together, the Saami languages form a dialect continuum; Pite Saami is therefore most closely related to Lule Saami and Ume Saami, the two languages spoken directly to its north and south, respectively. 1 Figure 1.1 on the following page shows the Saami sub-branch of the Uralic family tree.  nicipality it is spoken in. The endonym is bidumsáme giella or bisumsáme giella, which literally mean 'Pite Saami's language', as shown in example (1).
(1) Bidumsáme bidum+sáme Pite+Saami\gen.sg giella giella language\nom.sg 'the Pite Saami language' [pit080621.41m10s]e Nonetheless, speakers of Pite Saami (as is true for speakers of other Saami languages) generally refer to themselves and the individual language they speak simply as 'Saami', without further specification. This is likely the reason why there is no consensus concerning the endonym, which has two forms (see above); indeed, some speakers are quite unsure that an endonym exists at all.
Germanic cultures have often referred to Saami peoples using the exonym Lapps (cf. the place name Lapland), and thus the language has also been referred to as Pite Lappish in the past. This term is no longer considered respectful by many Pite Saami individuals; the name Saami is preferred, as it is borrowed from the endonym sábme or sáme. Nonetheless, a number of speakers I have worked with still refer to their own language as lapska when speaking Swedish.

Geography
The Saami languages are spoken in an area traditionally referred to as Sápmi; this covers a territory stretching from south-central Norway and central Sweden, across northern Norway, Sweden and Finland and over most of the Kola Peninsula in the Russian Federation, as illustrated in Figure 1.2 on the next page.
There is no official geographic or political unit defining any Pite Saami linguistic or ethnic area, but the individuals (including both speakers and non-speakers) traditionally spoken, with Pite Saami shaded in (borrowed from Bull et al. 2007, with permission) I have met, worked with or heard about who consider themselves to be Pite Saami (regardless of language abilities) all come from an area based roughly on the Arjeplog municipality 3 in Swedish Lapland and bordering areas in Norway. On the Swedish side, this has traditionally been referred to as Pite lappmark 'Pite Saami territory'. For instance, Ruong, himself a native speaker of Pite Saami, claims that the "most genuine form" (Ruong 1943: iii; my translation) of the Pite Saami language is spoken by members of the Luokta-Mavas sameby, 4 whose summer reindeer grazing lands are located along the headwaters of the Pite River, and by settled Saamis in the same area. 5 Manker's ethnography of the Saami populations in the Swedish mountains (Manker 1947) outlines the three samebyar 3 Note that Arjeplog municipality (kommun in Swedish) refers to the larger administrative district of ca. 15,000 km 2 , while the town of Arjeplog is the main village in the municipality. 4 A sameby (Swedish, literally 'Saami village'; samebyar in plural) is a group of reindeer herding families who tend their reindeer together in the same territory. 5 However, speakers from Semisjaur-Njarg sameby to the south were also included as sources in Ruong (1945), indicating that Ruong was clearly aware of the difficulty of drawing distinct language borders in the Saami dialect continuum. He indicates that some areas on the north side of the Pite River drainage speak Lule Saami, while speakers along the Skellefte River drainage are more under the "influence of Southern Saami" (Ruong 1945: iii; my translation).

Introduction
Luokta-Mavas, Semisjaur-Njarg and Svaipa as part of Pite lappmark. 6 Sammallahti (1998: 22) corroborates this, and adds the forest sameby Ståkke to the list. Collinder and Bergsland also help delineate the southern border of Pite Saami territory. Collinder writes that the border "goes along the Pite River between the parishes of Jockmock [sic] and Arvidsjaur, and farther west through the parish of Arjeplog" (1960: 23), while Bergsland further specifies that Ume Saami is spoken by "the forest Lapps in southern Arjeplog […] and by the mountain Lapps in Sorsele" (1962: 27).
As for the Norwegian side, some Pite Saami reindeer herding families had their summer reindeer grazing lands in the Norwegian territory adjacent to the international border (cf. Manker 1947). The Finno-Ugrian scholar Eliel Lagercrantz worked with Pite Saami speakers whose families originated in the Arjeplog municipality but had resettled to the Beiarn area in Norway (cf. Lagercrantz 1926). Ethnic Pite Saami individuals still live in Norway, and are, for instance, still active in the local Pite Saami association there, Salten Pitesamisk Forening.
As a result, one can say that Pite Saami was traditionally spoken in an area spanning both sides of the Norwegian-Swedish border around the municipality of Arjeplog on the Swedish side and across the border into Saltdal and Beiarn municipalities in Norway. On the Swedish side, the Pite Saami area is essentially limited to the Pite River drainage above the waterfall at Storforsen, and the sections of the Skellefte River drainage from the town of Arjeplog and farther upriver. The map in Figure 1.3 on the facing page gives a rough idea of the traditional geographic area, which is the light area on the map. It is based on Lagercrantz (1926), Ruong (1943), Manker (1947), Bergsland (1962) and Sammallahti (1998), as well as on my own knowledge gained by discussing family histories with Pite Saami individuals.
My own research indicates that Pite Saami is currently still spoken by a few members of the Luokta-Mavas, Semisjaur-Njarg and Ståkke samebyar, as well as by settled Saami families from the same areas. Furthermore there are a few speakers from the Arjeplog municipality who have since moved to other areas outside of Arjeplog municipality, even as far away as southern Sweden. Ethnic Pite Saami individuals from Norway have indicated to me that the last Pite Saami speakers on the Norwegian side died several generations ago.

The state of the Pite Saami language
Traditionally, most Pite Saami families lived either as semi-nomadic reindeer herders or as sedentary farmers, fishers and hunters. Despite having always been in contact, these two groups lived very different lifestyles, and spoke Saami in relative isolation from one another. While reindeer herding Saami have often been a topic of Saami studies, the settled Saami are neglected or even actively prejudiced against for not being 'true' Saami since they do not herd reindeer. The data supporting the present study and gathered as part of the Pite Saami Documentation Project (cf. §1.2.2) originated from both groups.
By the eighteenth century, the Saami peoples had, in general, been converted from animistic polytheism to Christianity, marking the beginning of Saami assimilation to North Germanic culture (cf. Pulkkinen 2005). Strict policies in the mid-nineteenth century sent Pite Saami children to special 'nomad schools' where they were not allowed to speak Saami. Attending these schools also kept them away from their families and from regularly participating in Pite Saami life, while reinforcing the state's drive to exclusively promote Swedish culture and social values (cf. Valijärvi & Wilbur 2011).
Accompanying this shift, traditional realms of Pite Saami experience are slowly being left behind in favor of Swedish ones. With the introduction of modern conveniences, most Pite Saami individuals have moved to permanent dwellings in populated communities, particularly Arjeplog, thereby leaving behind traditional ways of life. Along with this demographic shift, they are also losing the need to carry out traditional occupations using the Pite Saami language. As a result, all Pite Saami speakers today speak Swedish fluently, and indeed use Swedish significantly more often in everyday life (cf. Valijärvi & Wilbur 2011).
A gradual reverse in political and social policies in Sweden over the last decades led initially to the acceptance and then to the active promotion of multiculturalism, particularly concerning the Saami people. This has helped to positively change attitudes towards Pite Saami identity. However, concerning the Pite Saami language and many traditional cultural realms, this change seems inadequate for revitalization. For instance, the Swedish government's minority language law from 2010 only applies the blanket term samiska ('the Saami language'), and, in doing so, completely disregards the reality that there are five different Saami languages in Sweden. Even within Sápmi, Pite Saami is threatened by larger Saami languages, in particular by North Saami, which has an active speech community of around 30,000 speakers, regular television, radio, print and internet resources, and first-language instruction in public school (cf. Salminen 2007: 209-211). The robust status of North Saami in addition to the lack of an 1.2 Linguistic documentation of Pite Saami officially recognized Pite Saami orthography allow local government agencies to conveniently support North Saami alone in fulfilling the language law's requirements; any positive effects on the Pite Saami language are essentially negated.
According to my own data collected from Pite Saami individuals during fieldwork, there are around 30 speakers left, 7 of around 2000 ethnic Pite Saami individuals (Krauss 1997: 24). With one exception, all speakers are older than 50. Based on my own observations, all speakers are fluent in the Swedish language, even if many of them did not learn Swedish until they began school. Indeed, Swedish dominates everyday life for most speakers today, particularly for those who do not work in reindeer husbandry, and for those not living in the traditional Pite Saami area. Only a few households (less than five) use Pite Saami on a regular basis at home or in family situations, and these are still involved in reindeer husbandry. In public realms, the language is rarely used today. For a more detailed description and analysis of the situation the language is currently in, see Valijärvi & Wilbur (2011).

Linguistic documentation of Pite Saami
The Pite Saami language has been the subject of academic research in the past; cf. Halász (1896), Lagercrantz (1926), Ruong (1943) and Lehtiranta (1992). This extant body of research was consulted in coming to terms with Pite Saami linguistic structures in creating the present description. However, in fulfilling the goal of describing the Pite Saami language as it is spoken in the early 21 st century, the Pite Saami Documentation Project corpus, collected from 2008 through 2013, is the main source of data. With this in mind, I consider this a corpus-based description, as opposed to a literature-based description, which would have attempted to incorporate these previous works to a much greater extent. Indeed, everything not specifically cited as coming from another source is based on my own analysis of the corpus. The current description will hopefully be seen as a supplement to previous work on Pite Saami, and together with previous studies, can facilitate exploring language change as exemplified by a severely endangered language.
In the following sections, previous studies on Pite Saami are summarized in §1.2.1. The current corpus and the documentation involved in creating it are described in §1.2.2. Then §1.2.3 provides a short guide to using this description.

Previous studies
In the late 19 th century the Hungarian scholar Ignácz Halász studied Pite Saami, and wrote several studies in Hungarian on the language. Lule-és Pite-lappmarki nyelvmutatvéanyok és szótár 8 (Halász 1885) contains the Gospel of Matthew in Lule Saami, and a combined Lule Saami/Pite Saami wordlist with translations in Hungarian and German. Népköltési gyüjtemény: a Pite Lappmark Arjepluogi egyházkerületéböl 9 (Halász 1893) consists of a significant text collection of short Pite Saami texts, transcribed using the traditional Finno-Ugrian transcription. Each text is translated as a whole into Hungarian. The majority of the texts are traditional narratives, and a few poems and songs are also transcribed and translated. Pite lappmarki szótár és nyelvtan 10 (Halász 1896) includes morphological paradigms and a Pite Saami wordlist with translations into Hungarian and German.
Sprachlehre des Westlappischen nach der Mundart von Arjeplog 11 (Lagercrantz 1926) is a grammar in German. It is based on three months of fieldwork during which Lagercrantz consulted three Pite Saami individuals who had settled in the Beiarn district in Norway but who were originally from the Arjeplog municipality. The book covers semantically driven descriptions of clause structures, a limited description of morphology, and an extended analysis of the phonological system, based on phonetic acoustic experiments.
Lappische Verbalableitung dargestellt auf Grundlage des Pitelappischen 12 (Ruong 1943) is the dissertation of Israel Ruong, who later became professor of Saami languages and culture at Uppsala universitet. Ruong's native language was Pite Saami; his dissertation, an elaborate categorization of Pite Saami verb derivations, is his only study specifically dealing with the Pite Saami language.
Arjeploginsaamen äänne-ja taivutusopin pääpiirteet 13 (Lehtiranta 1992) is a Finnish-language description of Pite Saami phonology and inflectional patterns, and is based on recordings, publications and archived materials on Pite Saami from 1950 and earlier. Some paradigms can be found at the end of the book, as well as several Pite Saami texts with phonetic and orthographic transcriptions, as well as Finnish translations. The phonetic transcriptions are presented in the transcription standard used in Finno-Ugrian studies.

Linguistic documentation of Pite Saami
Note that these studies deal with Pite Saami as it was spoken before 1950. Finno-Ugristian studies have traditionally dealt with historical-comparative studies, and have not always been concerned with the synchronic state of Saami languages. Indeed, the distance some scholars keep from the synchronic situation is highlighted by the erroneous claim by Lehtiranta that Pite Saami is "now extinct" (Lehtiranta 1992: English abstract). Consequently, the present study is the first extensive description of the Pite Saami language in English and for a general linguistic audience.
Since the present study is intended to be a synchronic description of the Pite Saami language as used in the early 21 st century and reflected by the Pite Saami Documentation Project corpus, the previous work mentioned above have played an indirect but important role in its creation. However, these works were referred to in detail particularly when the data from the corpus were not substantial enough to allow relatively certain conclusions to be drawn. Data based at least partly on sources other than the documentation corpus are clearly marked as such in this description. Specifically, the sections in Lagercrantz (1926) concerning phrasal and sentence-level syntactic phenomena in Part A 'Ausdruckslehre ' (pp. 19-99) were informative, while Part B 'Formenlehre ' (pp. 103-141), the paradigms throughout Halász (1896) as well as the paradigms in the appendix to Lehtiranta (1992: 150-166) were consulted regarding morphology. In writing Chapter 10 on derivational morphology, Ruong's thesis (particularly Chapters 6 through 40, which present his data) provided valuable insights into the variety and complexity of Pite Saami derivation from both morphological and semantic perspectives.
In addition to the academic linguistic studies mentioned above, a number of other texts exist concerning the Pite Saami language and its people. Valijärvi & Wilbur (2011) describe the current state of the Pite Saami language from the point of view of sociology of language. Sjaggo (2010) deals with the etymology of a selection of Pite Saami place names along the river Piteälven in the Arjeplog municipality. A large number of Pite Saami vuole 14 (songs in the Saami singing tradition of yoik) were recorded in the first half of the 20 th century. These can be found transcribed in a number of works: Tirén (1942) includes 139 transcriptions of Pite Saami melodies and lyrics, with German translations; Grundström & Väisänen (1958) have 93 songs by Jonas Eriksson Steggo in the form of transcribed melodies and lyrics, with translations in Swedish and German; Grundström & Smedeby (1963) provide 73 songs by a variety of Pite Saami individuals in the form of transcribed melodies and lyrics, also with translations in Swedish 1 Introduction and German. Wickman (1964) discusses a short Pite Saami text from a recording done in 1939 by Israel Ruong; the text is presented in three transcription standards (Finno-Ugrian close phonetic standard, the author's own phonemic transcription, and a modified North Saami orthography) and includes an English translation. Lars Rensund's books (1982Rensund's books ( , 1986 detail personal recollections by the author, himself a Pite Saami, and are interspersed with sentences and occasionally entire narratives in Pite Saami. Bylund (1956) provides an in-depth study of the colonization of Pite Saami territory by Swedish settlers up to the middle of the 19 th century. No educational materials, bible translations or other common texts exist in the Pite Saami language. With the exception of the works by Lars Rensund, most Pite Saami speakers today are not aware of any of the works mentioned above.

The Pite Saami Documentation Project corpus
The data forming the basis of the present study were collected as a part of the Pite Saami Documentation Project. This description is a direct result of that project, the main goal of which is the linguistic documentation of the Pite Saami language. The project has so far resulted in audio and video recordings documenting current language usage and grammatical structures and includes an archived corpus comprising 29,208 transcribed and translated Pite Saami words (as of early March 2014; cf. the Appendix for a list of recordings). From June 2008 until July 2011, the project was carried out by Joshua Wilbur at the Nordeuropa-Institut at Humboldt-Universität zu Berlin, with support from the Endangered Languages Documentation Programme (ELDP; a part of the Hans Rausing Endangered Languages Project, with financial support from the Arcadia foundation and hosted by the School for Oriental and African Studies (SOAS) at the University of London). A continuation of the project is underway in 2013 and 2014 at Albert-Ludwigs-Universität Freiburg thanks to generous continued funding from ELDP.
Current trends in documentary linguistics were taken into account. 15 Himmelmann's proposal that "a language documentation is a lasting multipurpose record of a language" (Himmelmann 2006: 1) is a defining motivation behind the project. Accordingly, the resulting documentation consists of a documentation corpus of a variety of linguistic genres, including Pite Saami situations potentially of interest to non-linguistic disciplines and to members of the Pite Saami 15 Cf. Bird & Simons (2003), Gippert et al. (2006), Woodbury (2011), Austin & Sallabank (2011), Grenoble & Furbee (2010), the book series Language Documentation and Description, among others.

Linguistic documentation of Pite Saami
language community themselves, as well as the present description. Initial results have been archived at five archives at international, national, regional and local levels: • The Endangered Languages Archive (ELAR) at the School for Oriental and African Studies in London; • The Language Archive at the Max Planck Institute for Psycholinguistics in Nijmegen; • Dialekt-, ortnamns och folkminnesarkivet i Umeå 16 (DAUM) in Umeå, Sweden; • Ájtte: Svenskt Fjäll-och Samemuseum 17 in Jokkmokk, Sweden; • Silvermuseet 18 in Arjeplog, Sweden.
Working with multiple archiving sites as well as having all data in a digital format help ensure accessibility to and longevity of the data. Access to the materials is available via the archives (in some cases, this is possible via the world wide web). Ideally, an archive should provide interested parties with access to archived materials, while respecting the privacy and the wishes of recording participants as necessary; with this in mind, access rights to the data related to any given session reflect the wishes of speakers involved in a specific session concerning availability to the linguistics and other scientific communities, the Pite Saami and greater Saami communities, and other individuals and groups in general. Furthermore, as part of a scientific endeavor, the claims made in this book about Pite Saami linguistic structures should be reproducible; with this in mind, the original data are available to the academic community via the Language Archive at the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands; see §1.2.3.2 for more details on accessing the archive.

Collection methods
Data were collected and recordings were transcribed during a total of 23 months at the field site in and around Arjeplog, Sweden, with the invaluable assistance 1 Introduction of a number of Pite Saami speakers. They were compensated for their time and effort with a modest consultant honorarium.
The documentation corpus consists of more than 55 hours of recordings covering a variety of genres; cf. the Appendix for a list of recordings, including an indication of genre and medium. As the morphological structure of Pite Saami words is quite complex, it was necessary to rely on elicitation techniques to gather a sufficient number of word forms for a wide variety of lexemes as a basis for morphological analyses. As a result, the majority of recordings (approximately 45 hours) consist of elicitation sessions intended to gather specific details concerning the structure of the language. These were often conducted using Swedish as a metalanguage, but Pite Saami was used whenever efficient and useful, and more frequently in recordings made later in the project. A variety of elicitation methods were used. To a large extent, elicitation sessions were conducted as translational interviews (particularly in early recordings to collect initial wordlists) and sentence completion (using both Swedish and Pite Saami triggers, mostly to complete morphology paradigms). However, other methods were used as well, such as vocabulary card ordering tasks to test syntactic structures, and tasks using toy blocks to gather data on spatial relations. Many non-elicitation linguistic situations were also recorded, covering genres such as conversations, explanations, narrations, performances, as well as songs and readings; in the current documentation corpus, such recordings comprise approximately 17,700 transcribed words. A few written texts were also collected to supplement recordings.
Each collection of materials 19 in the corpus corresponds to a recorded linguistic event. Each collection has a unique name based on the pattern: pitYYMMDDabc (pit = Pite Saami, YYMMDD = abbreviated date of recording with a two-digit year, abc = further disambiguation as needed). Recordings done for the project from 2012 onwards use the ISO 639-3 code sje as a prefix for session names instead of pit, and a four-digit year; e.g., sje20121014b. All digital files related to a certain collection are named based on this pattern.
In almost all cases, the following recording equipment, standards and software were used for documentation. A small number of deviations exist, and are indicated in the archived metadata for the relevant sessions.

Introduction
• transcription/annotation file in ELAN and in plain text format • metadata concerning the session in XML and plain text formats • metadata concerning the entire collection in XML In addition to the above files, collections may also include the following files: • video recording in MPEG-2 or MPEG-4 format • digital images in JPEG format

• other supplementary files
The transcription/annotation files are divided into numbered, utterance-based units and include at least a transcription of any Pite Saami language usage. A specific utterance can be referred to using the collection/session name and the utterance number. The Pite Saami original is translated into Swedish and/or English, and provided with linguistic glosses. Other comments are included as well, whenever deemed relevant or useful. Finally, in cases with code-switching, the language being used in a certain utterance, or part thereof, is indicated. Tiers in all ELAN files are organized hierarchically based on the template in Figure 1.4 for each speaking participant in a recording, plus a 'notes' tier for general comments.

Using this description
This description of the grammar and the accompanying documentation corpus are together intended to outline the structural features of the Pite Saami language and provide an empirical foundation for claims made in the present work. In addition, the documentation should give interested individuals (linguists, Saami individuals, etc.) the opportunity to explore other aspects of the language as well. Finally, it should also secure a record of the language as spoken by those Pite Saami individuals who likely comprise the last generations of speakers for future generations of ethnic Pite Saami individuals.
The following sections are meant to assist in utilizing both the description and the accompanying documentation corpus. The first section ( §1.2.3.1) deals with accountability and verifiability, while the second section ( §1.2.3.2) provides brief instructions on how to access data from the corpus. Then, §1.2.3.3 illustrates how examples from the corpus are presented. Finally, §1.2.3.4 covers orthographic considerations and includes a list of phonemes and the graphemes used to represent these. A list of symbols and abbreviations used in the present work is provided starting on page xiii.

Accountability and verifiability
Data in the present study are cited with a bracketed reference to the collection name followed by the specific utterance number or timecode within a recording. This should allow the reader to easily find the evidence cited and make his/her own judgement about the conclusions made. References to specific recordings marked with 'e' after the closing bracket indicate that the presented utterance was attained during an elicitation session. For instance, the reference [pit090930a.239]e refers to utterance number 239 on recording pit090930a, and indicates that this example is from an elicitation session.
In order to verify, scrutinize or otherwise review the actual primary data on which the present study is based, the data can be accessed via one of the archives listed in §1.2.2. Specific instructions are provided in the following section.

Accessing archived materials
The archivists at the IMDI archive located at the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands, have kindly agreed to host the Pite

Introduction
Saami Documentation Project spoken language corpus. For those interested in accessing the data in connection with the present description, I recommend using this archive. Pite Saami materials are available via the on-line IMDI-browser at http://corpus1.mpi.nl/ds/imdi_browser/ under the hierarchical node: Endangered Languages/Donated Corpora/Pan-Saami Language Archive/Pite.

Explaining examples
Examples from the corpus are numbered consecutively for easy reference, and consist of several lines of text. The spoken text is presented in italics in the initial line in the orthographic standard, followed by interpretive information in the form of a morpheme-by-morpheme breakdown in the second line and English glosses in the third line. Finally, the last line contains a free translation in English and a reference to the source recording in the corpus; this reference includes either the number of the specific utterance or its initial time code. 22 Examples are formatted as shown in Figure 1.5.

(no.) Pite Saami original text
Pite Saami text with morpheme boundari-es Morpheme-by-morpheme glossing 'free English translation' [source] In Chapter 2 on prosody and Chapter 3 on segmental phonology, many examples of individual words are provided to illustrate phonological aspects of Pite Saami. In most cases, the phonological representation and phonetic realization (both using IPA standards) and the orthographic representation (based on the current working version) are included, as well as a gloss and a reference to the source recording, as summarized in Figure 1.6 on the next page. In such examples, the phonemic representation also indicates any linear morpheme boundaries. The source recording for these words indicates the recording name and utterance number or time code in most cases. However, when only a four-digit number is present (e.g., [0457]), this indicates that the original recording is not from the documentation corpus, but from the Wordlist Project (cf. §1.2.3.4). The number refers to the record number of the word in the Wordlist Project's lexical . Some examples in Chapter 10 on derivational morphology are also from this lexical database, and are marked in the same way.

Orthographic considerations
At the time of writing, Pite Saami does not have an officially recognized orthography. In the past, adapted versions of the Swedish orthographic standards 23 and Lule Saami orthographies 24 have been used to write Pite Saami for a nontechnical, non-linguistic audience. However, from 2008 through 2011, Arjeplogs sameörening (the local Saami association in Arjeplog) received funding to complete a lexicographic project called Insamling av pitesamiska ord, 25 (cf. Bengtsson et al. 2011) and hereinafter referred to simply as 'the Wordlist Project'. One of the outcomes of the Wordlist Project was a working orthography for Pite Saami. I have attempted to adopt this working orthographic standard in writing Pite Saami data in this description and the transcriptions provided in the accompanying documentation. This orthography uses the Swedish alphabet and many Swedish sound-to-grapheme correspondences, but also resembles to some extent Lule Saami orthographic systems, particularly the most recent one as found in Korhonen (2005). To help understand the orthography, the correlations between sounds and graphemes are discussed here and listed in It is important to note that the standards used in this book are based on a working orthography, i.e., it is still subject to inconsistencies and potentially to further refinement. Moreover, the following description and the implementation of the orthography is based on my interpretation of the recurring patterns used by the Wordlist Project in developing the orthography proposal mentioned above. While I generally use the spellings found in the Wordlist Project's wordlist, some deviations may be found in this description and the accompanying transcriptions; these can be due to a variety of factors such as changes to the wordlist as it was developed, deviating analyses on my own behalf, or simple inconsistencies in spelling (a natural occurrence in the development of an orthography) in the source. However, I alone am ultimately responsible for the orthographic choices in the present work.
The unaspirated singleton plosive phonemes are normally represented by the graphemes <b>, <d> and <g>, except as a single word-final consonant, in which case <p>, <t> and <k> are used; this is illustrated for the alveolar plosive phoneme /t/ in (2) through (4).
In general, geminate consonants are written by doubling the relevant grapheme, as illustrated in (7) through (9).
If a preaspirated segment is preceded by a sonorant consonant segment, then the preaspiration itself is not marked by a grapheme of its own, as in (15) through (17).
For plosives, the preaspiration is evident nonetheless because the corresponding plain segments are spelled differently, using <b, d, g>. However, the current working version of the orthography does not provide a way to distinguish between preaspirated affricates preceded by a sonorant segment, e.g., (17), and plain affricates preceded by a sonorant segment. In spoken Pite Saami, when the copular and auxiliary verb lä is preceded by a word ending in an open syllable, it is often encliticized as l on that preceding word. To reflect this in the orthography, lä is then written as 'l immediately following the preceding word, as in (18)

Typological profile
Pite Saami is a Western Saami language in the Saamic branch of the Uralic language family. It is currently spoken by around thirty speakers in and around the Arjeplog municipality in Swedish Lapland. (Cf. §1.1 for more details on the current state of the language.) With the exception of a limited number of grammatical items, Pite Saami words consist minimally of one trochaic foot. All non-final odd syllables are stressed, with the initial syllable being most prominent. (Cf. Chapter 2.) There are 43 consonant phonemes and 9 vowel phonemes. With the exception of the glottal fricative /h/, there is a length distinction for all consonants (singleton and geminate pairs). There are both voiceless and preaspirated plosive and affricate phonemes. Geminates and preaspirated segments are restricted to footmedial position. Vowel length is only distinctive in open front position (/a/ and /aː/). (Cf. Chapter 3.) Linear morphology in Pite Saami is exclusively suffixing. However, grammatical categories are often expressed non-linearly as well. This can take the form of foot-internal consonant alternations, umlaut in the first vowel of the initial foot, and regressive vowel harmony between both vowels of a foot.

Typological profile
Nouns inflect for number and nine cases. Verbs inflect for person, number, tense and mood. Adjectives come in sets of attributive and predicative forms that are not regularly derivable from one another; attributive adjectives do not inflect, while predicative adjectives inflect for number. Number distinctions are limited to singular and plural for nouns, non-personal pronouns and predicative adjectives, but also exhibit a dual category in pronouns and in verb agreement morphology. (Cf. §4.1 for a brief introduction to Pite Saami morphology; details on inflectional morphology can be found throughout Chapters 5 through 9.) There are seven word classes (verbs, nominals, adjectivals, adverbs, postpositions, conjunctions and interjections); these can be distinguished by syntactic criteria as well as their behavior concerning inflectional morphology. Nouns, verbs, adjectives and adverbs can be derived using linear and/or non-linear morphological processes. (Cf. §4.2 for an overview of the various word forms; Chapter 5 on nouns; Chapter 6 on pronouns; Chapter 7 on adjectivals; Chapter 8 on verbs; Chapter 9 on the other word classes; Chapter 10 provides some examples for derivational morphology.) Nominal, adjectival, adverbial and postpositional phrases and the verb complex constitute the main components of Pite Saami clauses, and are covered in Chapter 11.
Pite Saami has nominative/accusative argument alignment. Basic clauses consist minimally of a single finite verb form and potentially further non-finite verb forms, as well as any arguments, complements and adjuncts. Copular clauses require the fully inflected copular verb. Negation is expressed by the fully inflected verb of negation in combination with a special non-finite form of the negated lexical verb. Polar interrogatives can be identified by a question marker, but this is exceptionally rare in current Pite Saami usage. Clause-level possession can be expressed using a transitive verb with the possessor as the subject and the possessum as the object, or using a copular phrase with the possessum as the subject and the possessor in an oblique case. Relativization uses a relative pronoun introducing a relative clause with a fully inflected finite verb; the relative pronoun is not restricted in the syntactic role it has in the relative clause. Constituent order is not determined syntactically, but by information structure. (Chapters 12 through 14 deal with clause-level syntax.) The Pite Saami language exhibits a number of features which are potentially remarkable from a general typological point of view, even if most of these features are not particularly unusual among the Saami languages. A selection of such features and the sections that deal with these are listed in Table 1.3 on the next page.

Prosody
This description of the phonology of Pite Saami begins with a discussion of prosodic structures before the segmental phonology is described. This choice of ordering is motivated by the important role that prosodic positions play in the distribution of phonemes (as well as in morphophonology). It is useful to first understand the prosodic structure of Pite Saami words before looking at their segmental composition here, and later to better understand morphophonology. While there are a number of monosyllabic functional words, all Pite Saami lexical forms and many functional words are minimally bisyllabic. The first two sections ( §2.1 and §2.2) describe the prosodic structures of these two groups of words. Then, utterance-level prosodic phenomena are dealt with in §2.3.

Monosyllabic word structure
While the majority of Pite Saami words are polysyllabic, a small set of functional words are monosyllabic. This set includes, for instance, some interjections, conjunctions and pronouns. These monosyllabic words consist of at least one vowel 1 and one consonant. This consonant can be in either onset or coda position; it is also possible for both consonant positions to be filled. Consonant clusters are licensed in coda position as well. The possible segmental structure templates for monosyllabic words are listed with examples in Table 2.1 on the following page.

Multisyllabic word structure
All lexical forms and a large number of functional words in Pite Saami are minimally bisyllabic. The smallest prosodic segmental structure attested for polysyllabic words is VCV 1 All vowel phonemes except /u͡ a/ are attested in monosyllabic words.

Prosody
but larger words are both possible and common, and expand upon this minimal foundation; examples are provided throughout the following discussion. Due to a number of phenomena, it is sensible to posit a phonological domain, which, in following basic principles of phonology and the prosodic hierarchy (cf. e.g., Dixon 2010: 280-283, Selkirk 1980, Hayes 1989, Nespor & Vogel 1986, I will refer to as a foot. A Pite Saami foot is trochaic (counting from left to right) and essentially bisyllabic. Multisyllabic words with an odd number of syllables thus have a final (unstressed) syllable which falls outside of the last foot. Whether such a final syllable should belong to the preceding foot or not is a theoretical question which will not be addressed here, but it should be noted that the segments in such syllables are subject to highly restrictive phonotactics compared to those clearly located within a trochaic foot. 2 Evidence for the foot as a domain can be found in prosodic (intonation, cf. §2.2.1; minimal size restrictions as described here), phonological (segmental restrictions, cf. Chapter 3) and morphophonological (stem alternations and vowel harmony, cf. §4.1.2) phenomena.

Word stress
The initial syllable (cf. §2.2.3 on syllabification) of a Pite Saami foot always receives main stress. All other foot-initial syllables receive secondary stress. If 2.2 Multisyllabic word structure a final syllable is odd, it does not receive any stress. As a result, the patterns of stressed and unstressed syllables presented in Figure 2.1 are attested in Pite Saami; Note that some recent borrowings from Swedish deviate from these structures by having an initial unstressed syllable, as they do in Swedish. For instance, the example in (1) is from Swedish departement.
'department\nom.sg' [3583] The acoustic correlates for stress seem to be intensity and pitch. Note that vowel length does not play a role in stress. Indeed, there are words with a short first vowel and a long second vowel that receive stress on the first syllable, as in the two examples in (2) and (3).

Prosody
However, more detailed analyses is required to fully describe the acoustic-phonetic behavior of word-level stress, including the difference between primary and secondary stress.

Relevant prosodic domains
Due to systematic restrictions on the distribution of a number of segments and consonant clusters as well as to the prosodic domains of morphophonological processes, it is useful to name and describe various prosodic positions for polysyllabic words. The domains themselves are described below, while the relevant phonological restrictions and morphophonological processes are described in the pertinent sections on consonant phonemes ( §3.1), vowel phonemes ( §3.2) and morphophonology ( §4.1.2). Only a very limited number of recent loan words do not adhere to this structure. The schema in Figure 2.2 shows these prosodic positions, and they are described further below. In the schema, only segments represented by bold capital letters are obligatory.
c o n s o n a n t c e n t e r V 2 C 2 V 3 C 3 foot Figure 2.2: Illustration of prosodic domains for segments and the foot; segments represented by C and V are obligatory, while c and v are not.
Note that Table 2.3 on page 31 provides examples of how a word's segments fill these positions; it may be useful to refer to this to better understand these prosodic domains.

Foot
A foot in Pite Saami is a prosodic unit consisting of a stressed syllable and the following unstressed syllable, and is thus trochaic. Every multi-syllabic Pite Saami word consists of at least one foot.

Foot onset
Foot onset position is the first consonant or consonant cluster of a foot. It is not obligatorily filled. In Saamic linguistics, this has typically been referred to as the 'initium ' (cf. Sammallahti 1998: 39).

V1
V1 is the first vowel of a foot, and is the peak of the stress-carrying syllable for the foot. It can be long or short, and can be a monophthong or a diphthong. The vowel in the final V1 position 3 of a word is the location for umlaut and ablaut/jharmony (cf. §4.1.2.2 and §4.1.2.3). In Saamic linguistics, this has been referred to as the 'vowel center ' (cf. Sammallahti 1998: 39).

The consonant center
The consonant center is the consonant or consonants that follow V1 (the initial vowel) and precede V2 (the second vowel), and essentially form the core of a foot. Every foot has a consonant center. The final consonant segment of the consonant center is the onset of the second syllable due to syllabification (cf. §2.2.3). The final consonant center of a word is the location for consonant gradation. The term 'consonant center' is commonly used in Saami linguistics (cf. Sammallahti 1998: 39).

V2
V2 is the second vowel of a foot. It never carries stress. Every foot has a vowel in this position. With the exception of the diphthong phoneme /u͡ a/, all vowel phonemes are attested here. In Saamic linguistics, this has been referred to as the 'latus ' (cf. Sammallahti 1998: 39).

C2
C2 is the consonant or consonants following V2. It is not obligatorily filled. If it is followed by a V3, then its final segment is resyllabified as the onset of the following syllable. In Saamic linguistics, this has also been referred to as the 'consonant margin ' (cf. Sammallahti 1998: 39).

V3
V3 is the unstressed vowel of any syllable following the final foot of a polysyllabic word form, and is thus the last syllable nucleus of a word (when present). Only a limited set of vowel phonemes can occur in this position. If the V3 position is filled, then there is always a consonant margin as well. In Saamic linguistics, this has been referred to as the 'vowel margin ' (cf. Sammallahti 1998: 39).

C3
C3 is a consonant or one of a limited set of consonant clusters (cf. §3.1.2.2) following V3 of a polysyllabic word, and is thus always word-final (when present). It is not obligatorily filled. If the C3 position is filled, then there is always a vowel in V3 and a consonant margin as well. In Saamic linguistics, this has been referred to as the 'finis ' (cf. Sammallahti 1998: 39). Table 2.3 on the next page provides several examples for Pite Saami polysyllabic words and how their segments fill the prosodic domains described above. Note that there are always segments in V1, the consonant center and V2, forming a sort of 'minimal core. ' Only the final foot of a word can be followed by a single, odd syllable with V3 and potentially C3 segments.

Discussion and examples
Similarly, only the final foot of a word is subject to morphophonological phenomena (cf. §4.1.2). For instance, sálbmagirrje 'book of psalms, hymnal' is a compound consisting of sálbma 'psalm' and girrje 'book'. It consists of two feet: sálbma-and -girrje. It is not possible to add another syllable (e.g., via suffixation) between these two feet because they belong to the same compound noun. Furthermore, the inflected form for acc.sg is sálbmagirjev, in which consonant gradation (weakening of rr to r) is only triggered in the second foot, even though the first foot undergoes gradation (weakening of the cluster /lpm/ to /lm/) in non-compound environments, cf. sálmav 'psalm-acc.sg'. This is illustrated by the word forms in Table 2.4 on the facing page.

Syllabification
The distribution of vowel phonemes between consonant phoneme slots patterns clearly, particularly with respect to intonation and the distribution of vowel phonemes. This, along with the sonority sequencing principle (cf. e.g., Selkirk 1984), indicates that vowels are the nuclei of Pite Saami syllables. However, the location 2.2 Multisyllabic word structure Because the consonant center has by far the widest variety of consonants and consonant combinations of any of the consonant positions, it is best to consider this position first. Although the consonant center spans the preceding and following syllabic nuclei, there is no solid phonotactic or phonological evidence for where the syllable boundary is located inside the consonant center. Table 2.5 on the next page lists the possible syllabification patterns for the consonant center.
Maximizing onsets, the patterns V.CCCV, VC.CCV and V.CCV would create Similarly, trying to maximize codas, the patterns VCCC.V and VCC.V would also result in highly unusual codas (such as /vkŋ/ or /vɲ/) unattested in any other coda positions. While the pattern VCC.CV would also create some otherwise unattested codas (such as /vt/ or /rk/), these are phonologically similar to attested word-final codas such as /st/ or /jk/ (also fricative+plosive and oral-sonorant+plosive, respectively). The patterns VC.CV, V.CV and VC.V result in onsets and codas which are not unusual. However, keeping in mind the far greater diversity of single consonant phonemes licensed in word-initial onset position compared to word-final coda position, syllabification favoring singleton onset consonants results in onsets and codas which most resemble word-initial onsets and word-final codas. Note that even then, onsets and codas in positions other than the consonant center form subsets of the possibilities in consonant center position (with the exception of a few non-native word-onset clusters). It is therefore most plausible that syllables are assigned a single consonant segment as an onset in syllabification. The examples in (4) through (9)  This syllabification preference for a single onset segment can be applied to syllable boundaries outside of the consonant center, as shown in (10) and (11).
'speak-inf' [1480] (11) 'happiness-com.sg' [4372] When the consonant center consists of a geminate phoneme (cf. §3.1.1.2), there is no phonological test which indicates where the syllable boundary is located; indeed, this syllable boundary is likely not relevant in Pite Saami phonology. With this in mind, a symbolic syllable boundary is postulated somewhere within a phonological geminate; this divides the geminate symbolically into two component parts and results in syllables conforming to a syllable template with a singleton in the onset. In the examples in (12)  If a geminate precedes another consonant segment in the consonant cluster, the syllabification border is after the geminate, as in (16): 'spoon\nom.sg' [0190] When the final consonant in the consonant center is preaspirated, there is again no test for the location of a syllable boundary, as with geminates. With this in mind, a symbolic boundary is posited between the realization of preaspiration (cf. §3.1.1.1) and the rest of the preaspirated segment, which results in syllables conforming to a syllable template with a simgleton in the onset. The examples in (17) and (18) show a preaspirated plosive and affricate, respectively, with a syllable boundary indicated between the glottal fricative (preaspiration) and the stop or affricate component. 'house\nom.sg' [0416] The same is the case when a geminate precedes a preaspirated segment, as in the example in (19).
'church\nom.sg' [0640] Note, however, as pointed out above, the actual position of syllable boundaries in Pite Saami does not seem to be relevant in other areas of prosody. For this reason, the consonant center is a preferable prosodic domain to consider when describing prosody and phonotactics, and thus is referred to regularly in the following descriptions.

A note on syllables and feet
With the above description on stress in polysyllabic word structure in mind, it should become clear that syllables are only relevant for bearing word stress and creating feet, while feet form a relevant unit on several levels (prosodic, phonological, morphophonological). As a result, it could be more useful to rephrase the 'bisyllabic minimal word structure' as 'obligatory footedness' for Pite Saami lexical items and many functional words. Furthermore, the choice of the term 'foot' to describe this minimal size requirement may not be ideal because the edges of the Pite Saami foot are quite irrelevant to morphophonological processes. Instead, the V1+Consonant-Center+V2 core is a vital domain for morphophonology, while segments at the edges are not relevant. Perhaps a better descriptive term would be 'minimal core'.

Utterance-level prosody 2.3.1 Intonation in utterances
While the following observations are of a preliminary nature, and a more thorough study must be left for future investigation, the relative intensity of stressed syllables in declarative utterances in Pite Saami tends to decrease towards the end of the utterance, with the final stressed lexical item, and particularly the final syllable, being realized with noticeably lower intensity than the beginning syllables. As an example, the waveform and intensity trace for the utterance glossed in (20) are provided in Figure 2 gillgi-jmä will-1pl.pst sarvajd sarva-jd reindeer.bull-acc.pl njuovvat njuovva-t slaughter-inf 'It is autumn and we will slaughter the reindeer bulls. ' [pit090826.003] Here, the initial syllable nucleus of the first lexical item in the sentence tjakttja has an intensity of 69.7 dB. The other lexical items hover between 63.5 dB and 70.3 2 Prosody dB. The final lexical item njuovvat begins at 64 dB on the initial syllable nucleus, and drops abruptly to 50 dB on the final syllable nucleus.

Utterance-final weakening
The final two or three syllables of a declarative utterance in Pite Saami can be weakened as a way to mark the end of an utterance. 4 This weakening is typically realized by completely devoicing the final one, two or three syllables, often to the point that these are essentially whispered. Alternatively, this can be realized as creaky voice instead of voicelessness. For instance, in the example utterance depicted in the waveform in Figure 2.4 and transcribed in (21), devoicing occurs in the final lexical word giesev 'summer', which contains the last two syllables of the utterance.  The lack of energy in the waveform corresponding to giesev shows clearly that the word is weakened significantly compared to the rest of the utterance. Note that even the vowels are completely devoiced.
It is not clear what triggers this devoicing, and future study, particularly looking at the possibility of it being a turn marker in discourse, is needed.

Consonants
The consonant phoneme inventory of Pite Saami can be found in Figure 3.1. There are plain and preaspirated phonemes for all plosive and affricate positions, and geminate and singleton pairs for all categories. Preaspirated, geminate and preaspirated geminate phonemes are restricted to the consonant center position.  A description of the consonant phonemes and the distribution of their relevant allophones can be found in §3.1.1 for each mode of articulation. This is followed by a discussion of consonant clusters. For the sake of clarity, the term postaspiration will be used here to refer to what is commonly referred to simply as aspiration; this decision also emphasizes the contrast to preaspiration, which is, in fact, more relevant for Pite Saami than postaspiration.

Preaspiration
In Pite Saami, preaspirated 1 phonemes only occur in consonant center position. While preaspirated phonemes can be plosives or affricates, the phenomenon that goes along with them is essentially the same. The period of aspiration, i.e., the voicelessness preceding the formation of the oral closure, is realized in different ways and depends on the preceding segment. If the preceding segment is a voiced continuant consonant, then the final part of that segment is devoiced.
While a minimal contrast between a voiceless obstruent preceding a preaspirated consonant phoneme and a voiceless obstruent preceding a plain counterpart phoneme (e.g., /st/ vs. /sʰt/) is theoretically possible, this cannot be detected because such a consonant cannot be devoiced as it is already voiceless (e.g., /st/→[st] and /sʰt/→[st]). When following a high front vowel /i/, preaspiration is realized as a voiceless palatal fricative [ç]. In all other cases, preaspiration is a voiceless glottal fricative [h]. This is summarized in Table 3.1.

Geminates
The occurrence of geminate consonants is restricted to the consonant center. As illustrated in Table 3.1 on the preceding page, only the glottal fricative /h/ does

Consonants
not have a comparable geminate phoneme. Geminate segments are realized with a longer overall duration than the corresponding singleton phonemes. This observation is based not only on speakers' observations that such sounds are 'longer', but also on my own observations and analyses, including an acoustic-phonetic comparison of duration in geminate phonemes; cf. §3.1.1.3.5 for a detailed comparison non-voiced plosive durations in the consonant center. For plosives and affricates, only one stop closure is formed, and the overall duration of the stop closure is longer than the duration of a corresponding single consonant. Some examples are provided in (1) through (5).
'cervical.vertebra\nom.sg' [3771] Due to the morpheme boundary separating such segments, it is clear that this is not a case of a geminate phoneme, even if the realization may resemble that of a geminate.

Plosives
The plosive series in Pite Saami consists of the phonemes and their phonetic realizations shown in Table 3.2 on the next page. The distribution of the allophones will be discussed here. As all three relevant places of articulation behave in much the same way, the various manners of articulation for each place will be treated together.

Voiceless singleton plosives
The segments /p t k/ are bilabial, alveolar and velar (respectively) voiceless singleton plosive phonemes. The voiceless singleton plosives can occur in all prosodic consonant positions and are subject to allophonic variation, depending on the prosodic environment. In syllable-onset position, a plain (unaspirated) voiceless plosive [p t k] is produced, as seen in examples (26) through (32). The plain voiceless singleton pronunciations [p t k] are also found in consonant clusters in word-onset position, an environment usually found in recent and older loan words from (North) Germanic, as in examples (33) and (34).

Comparison of non-voiced plosive durations in the consonant center
While a thorough study of length phenomena in Pite Saami is beyond the scope of the current study, it is worth noting the actual duration which the various plosive phonemes are realized with in the consonant center. Specifically, a plain geminate plosive and a singleton preaspirated plosive have approximately the same duration for the period of stop closure (around 300ms), and are at least 100ms longer than plain singleton plosives and 100ms shorter than a preaspirated geminate plosive. In this respect, plain geminate plosives and preaspirated singleton phonemes seem to group together concerning stop closure duration. Table 3.3 on the following page shows some (near) minimal sets and duration measurements as a comparison. However, this alignment seems to be irrelevant phonologically.

Fricatives
The fricative series in Pite Saami consists of the phonemes and their phonetic realizations shown in Table 3.5 on the facing page.
'ford\nom.sg' [4332] (77) /nevres/ nievres [ni͡ evres] 'bad' [5101] For some speakers from the north-eastern parts of Pite Saami territory, /ʃ/ is also possible word-finally because the diminutive suffix is sometimes /-ʃ/(instead of /-ʧ/); however there are not enough data in the corpus to determine when the diminutive suffix is /-ʃ/. Secondly, a glottal fricative [h] as the sole consonant in word-final position is possible, but it seems to be limited to certain morphological conditions in contemporary Pite Saami, and only realized by some speakers, and then inconsistently. However, some of the literature describing older stages of Pite Saami 9 indicates that at a previous stage of the language, a word-final /h/ was obligatory when it had morphological status. Two examples for this variation can be seen in (79) and (80)
3.1.1.5.4 Dialect variation and the historical voiced dental fricative A number of Pite Saami lexemes historically featured a voiced dental fricative *ð in Proto-Saamic. These items are subject to variation in the corresponding synchronic phoneme across Pite Saami territory. Specifically, Proto-Saamic *ð can correspond to a singleton or geminate alveolar voiceless plosive /t tː/, alveolar trill /r rː/, or voiced dental fricative /ð ðː/; the selection of phoneme varies from speaker to speaker. The alveolar plosives and trills /t tː r rː/ are realized as described in §3. , respectively. 11 Phonemes subject to this variation are only found in the consonant center. To illustrate this, the phonemic variation in the word for stone, which goes back to Proto-Saamic *keaðɢē (Sammallahti 1998: 243), is presented in Table 3.6. Generally speaking, the phoneme /t/ is found on the northern side and the phoneme /r/ on the southern side, although the borders are not absolutely clear. The phoneme /ð/ is least common, and seems to only be found in the speech of the eldest speakers. Speakers are quite aware of this variation. In the current working version of the Pite Saami orthography, the grapheme <r> has been chosen to represent all three variants (thus the word for stone is spelled gärrge), although spellings using the grapheme <d> or even <ð> may be used as well.

Nasals
The nasal series in Pite Saami consists of the phones and their phonetic realizations shown in Table 3.7 on the next page. The distribution of the allophones will be discussed here, as well as in §3.1.1.8 concerning the devoiced allophones.
3.1.1.6.1 Singleton nasal consonants The segments /m n ɲ ŋ/ are singleton bilabial, alveolar, palatal and velar nasal consonant phonemes. They can be found in onset and coda positions, with the exception of the velar nasal, which cannot 11 The phonological system of the few speakers who have the voiced dental fricatives /ð ðː/ actually has two more phonemes than the systems of speakers with /t tː/ or /r rː/, as these latter four phonemes are already present in the phonology of all speakers. Note that, because /ð/ is very uncommon and not a feature that all speakers share, it is not included in the consonant inventory presented in Table 3.1 on page 37.

Oral sonorants
Pite Saami has three oral sonorant phonemes; because their behavior is very similar, they will be described together in the rest of this section. Their phonetic realizations are shown in Table 3.8 on the facing page, as well as in §3.1.1.8 concerning the devoiced allophones.  (109) and (110).

Sonorants and preaspiration
All sonorant phonemes become devoiced towards the end of their realization when preceding a preaspirated plosive or affricate. 13 Since preaspiration is limited to the consonant center, this devoicing is (with the exception of word-final devoiced /r/) also limited to the consonant center. Some near minimal pairs are listed in (124)  These examples all show a sonorant preceding a preaspirated plosive; note that a preaspirated affricate triggers the same devoicing in the preceding sonorant.

Consonant clusters
In Pite Saami, it is frequently the case that up to three consonants can occur consecutively, particularly in the consonant center. Because syllabification does not cross word boundaries, consonant clusters in word-initial and word-final position are necessarily tautosyllabic. However, word-internally, syllabification of the final consonant as a syllable onset (cf. §2.2.3 on syllabification) creates a syllable boundary within a group of consecutive consonants. There are two ways of approaching such word-internal consecutive consonant groups: on the one hand, one can consider the syllable boundary to be a significant fissure dividing such a consonant grouping into two units, and then only study any tautosyllabic consonant clusters that result. On the other hand, one can disregard any syllable boundaries, and thus treat any consonant groupings, even those spanning a syllable boundary (heterosyllabic consonant clusters), as a unit. In determining whether syllable boundaries are a meaningful part of Pite Saami phonotactics, a discussion of the inventories for both tautosyllabic and heterosyllabic consonant clusters is provided below. In the following, tautosyllabic consonant clusters will be described first, before moving on to heterosyllabic consonant groupings. Note that this does not include consecutive consonants which arise in compounding at an internal rootboundary.

Consonant clusters in syllable onset position
In syllable onset position, 21 CCs and 2 CCCs are attested, as listed in Table 3.9  and Table 3.10 on the next page; all are in word-initial position. 14 Words with an 3.1 Consonants onset cluster tend to be of either unknown or of Germanic origin, which helps explain why eleven of the word-initial CCs and both of the CCCs would not be attested in word-internal onsets even if syllabification allowed tautosyllabic consonant clusters word-internally. Table 3.9: Bipartite consonant clusters in syllable onset position C 1 C 2 attested CCs plosive + sonorant pr, pl, tr, kn, kr, kl sibilant + obstruent sp, st, sk, sv, ʃk, ʃv fricative + sonorant fr, fl, sm, sn, sɲ, sl, ʃm, ʃɲ, ʃl Table 3.10: Tripartite consonant clusters in syllable onset position C 1 C 2 C 3 attested CCCs s + plosive + r str, skr

Consonant clusters in syllable coda position
Because syllabification results in syllable onsets of a single consonant segment (cf. §2.2.3), only the coda of the initial syllable can host tautosyllabic consonant clusters in the consonant center. An inventory of these CCs is provided in Table 3.11. Table 3.11: Tautosyllabic CC clusters in the consonant center (all in coda position) C 1 C 2 attested CCs fricative + plosive vt, vk oral sonorant + plosive rp, lp, jp, rt, lt, jt, rk, lk Syllable codas in word-final position are more restrictive, as only three CCs and one CCC are frequently found in this position, while four other CCs are found in a very limited set of words; the attested clusters in word-final syllable codas are listed in Table 3.12 on the next page. The regularly occurring word-final coda clusters are quite common suffixes. 15 The clusters /rt rm lm jk/ are limited to a single, seemingly native lexical item each, but there are not enough data at this point to make any further conclusions.

Heterosyllabic consonant clusters in the consonant center
The inventories of tautosyllabic consonant clusters in various word positions detailed above are lacking any regularity concerning position within syllable structure (onset or coda). In other words, the sets of coda clusters licensed wordinternally only overlap with the coda clusters licensed word-finally to a very limited extent. Specifically, only the clusters /jt lt rt/ are attested both wordinternally and word-finally, while all the other clusters are unique to either wordinternal or to word-final position. Furthermore, the relatively large number of word-initial consonant clusters, but complete lack of consonant clusters in other syllable-onset positions word-internally is also asymmetrical. These facts indicate that perhaps a different approach to explaining the data would be more fruitful.
Keeping the above in mind, as well as the exceptional role that the consonant center plays in morphophonology (consonant gradation, cf. §4.1.2.1) and phonotactics (geminates, preaspiration, overall length), an inventory of the possible heterosyllabic consonant clusters, e.g., disregarding syllable boundaries, that occur in the consonant center as a unit proves more insightful in describing Pite Saami phonology.
In addition to the 21 geminate consonants that can occur alone in the consonant center, there are a total of 213 heterosyllabic CCs attested in the consonant center. Table 3.13 on the facing page lists the 197 heterosyllabic CCs with either two singleton consonants or a geminate consonant and a singleton consonant. 15 The suffixes -st elat.sg, -jst elat.sg, -jt acc.pl form an integral part of any noun paradigm.
The suffix -lt is limited to a handful of directional particles and may be an old case suffix. It should also be noted that Pite Saami speakers from the northern side of Pite Saami territory use -s and -js for elative case marking. Most combinations of various natural classes are found; however, it is striking that a nasal as the first element can only have an obstruent as the second element. It is also noteworthy that a single oral sonorant plus a nasal is attested, but no double oral sonorant plus a nasal.

Consonants
Turning to heterosyllabic consonant clusters with three members (tripartite CCs), there are 16 attested in the consonant center; these clusters are listed in Table 3.14 on the next page. The heterosyllabic consonant clusters in the first two rows of this table are fairly common, and correspond paradigmatically to consonant clusters lacking the plosive but with only the fricative and sonorant (see above on heterosyllabic CCs and §4.1.2.1 on consonant gradation). The other three tripartite consonant clusters /jst mst rtm/ are only attested in one or two Table 3.14: Tripartite consonant clusters in the consonant center C 1 C 2 C 3 possible CCCs fricative + plosive + sonorant vtn, vtɲ, vkŋ sonorant + plosive + sonorant rpm, lpm, jpm, rtn, ltn, jtn, rtɲ, ltɲ, rkŋ, lkŋ other limited CCCs jst, mst, rtm words each, 16 and there are not enough data to reach any further conclusions at this point. Due to the morphophonological process of consonant gradation, which features paradigmatic stem allomorphy characterized by quantitative alternations in the consonant center in many cases (cf. §4.1.2.1), almost all of the heterosyllabic CCs can be grouped into short∼long pairs, e.g., /pt∼pːt/ or /jʰʦ∼jːʰʦ/. There are only 16 heterosyllabic CCs which do not seem to have a corresponding quantitative partner; for reasons explained below, it is useful to divide these into the two groups listed in (132): (132) Group A: /vn vɲ vŋ lm ln lɲ lŋ rm rn rŋ jm jn jŋ/ Group B: /sŋ ʃm ɲk/ Members of the first and larger group all have a corresponding morphophonemic partner, but this corresponding partner is a consonant cluster consisting of three consonant segments, and differs qualitatively as well. Specifically, tripartite CCs consisting of /v l r j/ followed by a plosive+sonorant pair correspond to Group A (those lacking the plosive element of the relevant tripartite CC); these pairings are listed in Table 3.15 on the facing page. The remaining heterosyllabic CCs /sŋ ʃm ŋk/(Group B) seem to lack a quantitative partner. It is likely that the corresponding long CCs /sːŋ ʃːm ŋːk/ would be acceptable since CCs with very similar phonemic structures in quantitative pairs exist. However, a lack of data at this point prevents this from being ascertained for certain.
It is worth noting that the only consonant cluster which occurs in a wordmedial position other than the consonant center is /st/ from the suffix -st-, a derivational morpheme which derives a verb, e.g., basestit 'to fry quickly' (cf. basset 'to fry'). 17 16 Recent loan words from Swedish may also contain tripartite CCs not found elsewhere in Pite Saami, e.g., kɔnstɔ 'art' <Swedish konst. 17 Cf. §10.2.2 for more on this verbalizer. Ultimately, this massive inventory of consonants and consonant clusters in the middle of words can only really make sense when one understands the extent to which morphology is expressed in this position, as becomes clear in Chapters 4 through 10.

Vowels
Pite Saami has eight monophthong vowel phonemes and one diphthong vowel phoneme. These are listed in Figure 3.2, with the monophthongs in the vowel chart on the left, and the diphthong on the right.  A discussion of these phonemes and the distribution of the relevant allophones follows. Note that there is a short open front vowel /a/ and a long open front vowel /aː/; for the latter case, length is marked with a triangular colon <aː> in phonemic transcription, as according to IPA standards, and with an acute accent <á> when represented in orthography.
The monophthong phonemes /a aː ɛ i u ɔ/ are realized as monophthongs in all cases. The monophthong phonemes /e/ and /o/ are realized as slight diphthongs 3 Segmental phonology in V1 position, namely as [i͡ e] and [u͡ o], respectively. They can be very short in duration. There is a relatively minimal difference between the beginning and ending positions of each of the [i͡ e] and [u͡ o] phones in the oral cavity, a closeness which is even reflected in inconsistencies in the working version of the Pite Saami orthography: both <e> and <ie> are used for /e/, and <o> and <uo> for /o/.
The diphthong phoneme /u͡ a/ is restricted to V1 position. Due to vowel harmony, /u͡ a/ can be realized as [ʊ͡ ɛ] or [ʊ͡ a], and needs not be particularly long in duration.
Note that farther north within the Pite Saami language territory, monophthong phonemes are often realized as slight diphthongs. This is closer to Lule Saami within the Saami dialect continuum; indeed, many Lule Saami monophthong vowel phonemes are realized as slight diphthongs (cf., e.g., Spiik 1989: 11).
Because there is no significant difference in behavior between monophthongs and diphthongs which would justify treating them separately, all nine vowel phonemes are dealt with together in the following section.

Vowel phonemes and allophonic variations
The following sections describe the Pite Saami vowel phonemes and their allophonic realizations in the three vowel positions V1, V2 and V3. Table 3.16 on the facing page summarizes the distribution of the vowel phonemes in these three prosodic vowel slots. All vowel phonemes are licensed in V1, and all except /ɛ/ and /u͡ a/ occur in V2. 18 However, V3 position is the most restrictive and allows only /i ɛ a u/. This distribution reflects the fact that V1 is the most prosodically relevant slot, and V3 the least significant.

Close front high vowel
The segment /i/ is a close front high vowel. In V1 position, it is not realized with a very narrow oral cavity, but closer to [ɪ], while in V2 and V3 position it is even less close and essentially [ɪ], and tends to be shorter in duration. When a palatal approximant /j/ immediately follows, this triggers a slight raising of /i/ so that it is closer to [i]. Some examples are found in (133) through (136). 'decrease-inf' [0555] 18 The phoneme /ɛ/ only occurs in a limited phonological context in V2 of grammatical words, and is therefore in parentheses in Table 3.16.; cf. §3.2.1.3.

Close-mid front vowel
The segment /e/ is a close-mid front vowel. In V1 position, it is realized as a slight diphthong [i͡ e], while in V2 position it is normally a monophthong [e]. It is not attested in V3 position. 19 Some examples are found in (137)  3 Segmental phonology

Open-mid front vowel
The segment /ɛ/ is a open-mid front vowel. It normally occurs in V1 position. Examples can be found in (140) and (141).
'wind\nom.sg' [2302] (141) 'ice.fishing.line\nom.sg' [2754] The phoneme /ɛ/ can also be found in V2 position, but this is limited to grammatical words and it is never followed by a final consonant. 20 In such cases, /ɛ/ is realized slightly more open than when in V1 position. Examples can be found in (142) through (145). 'in.order.to' [0014] 'now' [2303] (145) /sinːɛ/ sinne [sɪnːɛ] 'inside' [pit080702b.144] In a few recent loan words from Swedish which are originally French or Latin loans words in Swedish that have retained their second syllable stress, /ɛ/ can occur as the vowel of the second syllable. However, as this is the stressed syllable in such cases, it is, from a prosodic perspective, still in the V1 position of a normal Pite Saami trochaic foot. Two examples are provided in (146) and (147).

Vowels
Due to the verbal suffixes -jmä '1pl.pst' and -jdä '2pl.pst', verbs with a bisyllabic stem in these forms thus feature /ɛ/ in the third and final syllable, as in (148) and (149). Note also that the current working orthography inconsistently spells /ɛ/ as <e> or <ä> in these suffixes.

Open-mid back vowel
The segment /ɔ/ is a open-mid back vowel. It can occur in V1 or V2 position. However, if it is in V2 position, then V1 is also /ɔ/. Examples can be found in (159) through (163).

Close back to open front vowel
The segment /u͡ a/ is a diphthong which begins as a close but slightly centralized back vowel and opens to an open front vowel [ʊ͡ a] in most cases. However, the 3 Segmental phonology vowel in V2 position can trigger vowel harmony that slightly closes the end position of the oral closure so that it is realized as [ʊ͡ ɛ], 23 but the triggering vowels vary between Pite Saami dialects. For southern dialects, only a close /i/ or closemid vowel /e/ in V2 position can trigger this harmony. In northern dialects, an open front /a/ in V2 position 24 also triggers this vowel harmony. A few cognate pairs are provided in Table 3.17. Note that this vowel harmony is triggered by a purely phonological context, as opposed to the vowel harmony described in §4.1.2.3 in the chapter on morphology, which is triggered morphologically.

Epenthetic schwa
In a small number of Pite Saami words, a vowel is inserted between two nonhomorganic consonants in the consonant center position. This centralized vowel is exceptionally short in duration, and is transcribed here with a superscript schwa [ᵊ]. Waveforms for the two examples provided in (174) and (175) are found in Figure 3.3 on the facing page.
'liver\nom.sg' [1403] (175) 'candy\nom.pl' [1277] 3.  In both cases, the epenthetic schwa is clearly linked to more energy, and it stands out from the surrounding consonants. These waveforms also make visible the shorter duration of the epenthetic schwa (59ms and 47ms, respectively) compared to the other vowels (the shortest of which is 108ms).
Speakers are rarely conscious of this vowel, and it is not reflected in the orthography. In neighboring Lule Saami, a similar epenthetic vowel exists and is predictable based on the prosodic and phonological structure of a word (cf. Spiik 1989: 14-15). It therefore seems likely that this epenthetic schwa is not phonemic in Pite Saami either. However, more data are needed to confirm this and thoroughly describe its distribution. The fact that this epenthetic vowel seems to be significantly more prevalent in northern Pite Saami dialects complicates the situation further. The examples in (176) through (178) provide dialectal variants, with the more southern variant first (lacking the epenthetic vowel), and the more northern variant second (with the epenthetic schwa).

Morphological patterns and word classes
Morphology plays an essential role in Pite Saami, a highly synthetic language. Based on morphological patterning and in addition to syntactic criteria, seven word classes can be posited. §4.1 first provides an introduction to morphological phenomena in Pite Saami, before §4.2 summarizes the word classes.

Overview of morphology
A number of inflectional categories exist in Pite Saami; Table 4.1 provides a summary of inflectional categories relevant for each sub-category of the relevant word classes. Derivational morphology is commonly used to create nouns, verbs, and, to a lesser extent, adjectives and adverbs. Both derivational and inflectional mor-4 Morphological patterns and word classes phology manifest themselves linearly (by suffixing) or non-linearly, via consonant gradation, umlaut and/or vowel harmony. More often than not, linear and non-linear morphological phenomena are combined.
The present section only provides an overview of these various morphological phenomena, and is divided into §4.1.1 on linear morphology and §4.1.2 on nonlinear morphological processes. Because morphological behavior varies between the word classes, it is described in more detail individually in the relevant word class chapters.

Linear morphology
Concerning linearly separable morphology, Pite Saami is an exclusively suffixing language. Both inflectional and derivational suffixes exist. According to the general linear morphological structure of Pite Saami words, derivational suffixes attach to a root before inflectional suffixes occur on the resulting stem, as illustrated in (1).

Non-linear morphology (morphophonology)
There are three ways in which non-linear morphology can be expressed in Pite Saami: • stem consonant alternations (consonant gradation) • stem vowel alternations in V1 position (umlaut) • regressive vowel harmony in V1 and V2 vowels These are triggered by a word's position within an inflectional paradigm, or in derivation. All inflectional non-linear morphology is restricted to the final foot of a given word, while derivational non-linear morphology can also occur in a non-ultimate foot. Non-linear processes may apply simultaneously. The following sections describe these phenomena in more detail: §4.

Consonant gradation
The term consonant gradation 2 refers to regular alternations of the consonant phonemes in the consonant center of the final foot of a word. 3 All consonant phonemes in the consonant center are included in the present classification of such alternations. These alternations come in pairs of stem allomorphs that differ quantitatively and/or qualitatively. Alternations can be between a preaspirated and the corresponding non-aspirated consonant (ʰx-x), a geminate consonant and the corresponding singleton consonant (xː-x), a gemi-nate+singleton and the corresponding singleton+singleton (xːy-xy), two singletons and only the latter singleton (xy-y), and three singletons and only the initial and final singleton (xyz-xz). 4 These patterns and the attested alternations are provided in Table 4.2 on the following page. The term strong grade (abbreviated 'str') is used to refer to the form with preaspiration, a geminate or more consonant segments than the corresponding form. Likewise, the term weak grade (abbreviated 'wk') refers to the form lacking a preaspirated or geminate consonant, or having fewer consonant segments, respectively. To facilitate reading Table 4.2, the attested alternations for each pattern are organized by alternations with a non-aspirated element first, followed by patterns with at least one preaspirated element. Furthermore, the individual lists of alternations are organized by mode of articulation and by the order set forth in the consonant phoneme inventory in Table 3.1 on page 37. Some examples illustrating this can be found below.
The minimal pair in (2) shows a consonant gradation alternation /ʰp-p/, which corresponds to the pattern ʰx-x.

Overview of morphology
consonant gradation alternations illustrated here are /tn-n/ and /jpm-jm/, respectively, and correspond to the patterns xy-y and xyz-xz.
Note that the geminate plosives and affricates /pː tː kː ʦː ʧː/ are lacking in Table 4.2 on the facing page for the pattern xː-x, although alternations such as pː-p could be expected. However, due to a lack of sufficient data and some conflicting data in the corpus, it is not entirely clear what the current status is for consonant gradation in words with a consonant center consisting solely of a geminate plosive or affricate. The fact that Pite Saami lacks consonant gradation in a limited number of contexts is one of the main differences to Lule Saami to the north, which does not lack consonant gradation, and Ume Saami to the south, which features consonant gradation even less frequently (cf. Sammallahti 1998: 21-23). The example in (7) illustrates a word clearly lacking consonant gradation in the corpus data, here with the geminate velar plosive /kː/.

(7)
/vaːkːe/ /vaːkːe/ (*/vaːke/) vágge vágge valley\nom.sg valley\nom.pl [pit110522] Corpus data also indicate that variation within the Pite Saami area complicates things. For instance, the adjective tjábbe 'beautiful' undergoes consonant gradation in the speech of speakers from the northern parts of Arjeplog, but does not for southern speakers, as illustrated in (8). For northern speakers, the gradation is realized as an alternation in voicing, and not length.
/ʧaːpːa/ /ʧaːpːe/ /ʧaːbːe/ southern northern tjábba tjábbe beautiful\attr beautiful\pred [pit110522, sje20131017] This being the case, some speakers from farther north may have further voiced plosive phonemes /bː dː gː/ that only occur in consonant gradation alternations with the corresponding unvoiced phonemes /pː tː kː/. However, it is not clear based on the corpus data how widespread this feature is, or if it also affects geminate affricates.
Finally, it is not clear from the corpus data what the status of the phonological contexts lacking consonant gradation mentioned in Sammallahti (1998: 21) (working from a historical perspective and with older data) is. Further research is needed to complete the picture, and variation within Pite Saami and possible effects of language attrition should also be taken into consideration.

Umlaut
The term umlaut refers to regular allomorphic alternations of the vowels in the V1 position of a stem. 5 The two umlaut patterns attested in the corpus are listed in Table 4.3.
These umlaut alternations are qualitative and not quantitative. These alternations are not triggered by the phonological environment, but instead morphologically. The allomorph /ɛ/ in the first pattern is found in the same paradigmatic slots for each inflectional class as /u͡ a/ 6 in the second pattern, just as the allomorphs /e/ and /o/ also correspond to the same paradigmatic slots. Word forms for bägge 'wind' in (9) and for buälldet 'burn' in (10)  For lexemes subject to consonant gradation, forms featuring /ɛ/ or /u͡ a/ are typically in the strong grade, while forms with /e/ or /o/ are normally in the weak grade; cf. the word forms in example (10).

Vowel harmony
The term vowel harmony (abbreviated 'VH') here refers to non-adjacent regressive phonological assimilation concerning the place of articulation of the V1 vowel of a stem in the context of certain V2 vowels. 7 Specifically, mid-high or high front vowels in V2 position in specific paradigmatic slots trigger raising of the vowel in V1 position. Because the paradigmatic slots that trigger vowel harmony differ between word classes and inflectional classes, and do not apply across the board, vowel harmony is not a purely phonological process, but morphophonological. Furthermore, the results of harmony on the same underlying vowel are inconsistent, and may be due to a word's membership is certain morphological classes concerning vowel harmony. However, future research must be conducted to come to a more thorough conclusion on this.
Verbs and nouns can be subject to vowel harmony, but the assimilation patterns vary both between these word classes and within them. Table 4.4 on the following page summarizes the various patterns and the word classes that they are attested in based on the current corpus.
The morphological categories that trigger vowel harmony also vary. This is the case not only between nouns and verbs (as these have different inflectional categories), but also between inflectional classes for verbs. These categories are presented in Table 4.5 on the next page.
Some examples for vowel harmony are provided here. In (11), an example is shown of vowel harmony in the Class Ie noun guolle 'fish', as it alternates between /o/ in the V1 vowel of the nom.sg form, and /u/ in the nom.pl form.
(11) /kole/ /kulij/ guole guli-j fish\nom.pl fish-gen.pl [pit110413a] In (12), an example of vowel harmony in the class II verb bassat 'wash' is provided. Here, a vowel harmony alternation between /a/ in the V1 vowel of the  Finally, (13) shows an example of vowel harmony in the class III verb buälldet 'burn'. Here, a vowel harmony alternation between /o/ in the V1 vowel of the 2sg.prs form, and /u/ in the 2sg.pst form is evident (in addition to a consonant gradation alternation).

Overview of word classes
See §5.3.1.2 for more details on vowel harmony in nouns, and §8.4.2.1 in verbs. Note that, for nouns, vowel harmony is also referred to as 'j-suffix vowel harmony'.

Overview of word classes
By characterizing the morphological and syntactic behavior of words in Pite Saami, and grouping such words based on that behavior, a total of seven word classes can be distinguished. These can be divided into two general categories containing generally open word classes and closed word classes, and are listed in Table 4.6. The specific syntactic criteria and inflectional categories defining these are summarized in Table 4.7 on the next page. Some word classes consist of two or more subclasses: nominals refer to nouns and pronouns (personal, demonstrative, reflexive, interrogative and relative), and adjectivals include both adjectives and numerals. Note that pronouns and numerals are closed subclasses belonging to open classes.
This categorization is intended to provide a broad starting point for classifying Pite Saami words; details for each word class can be found in the relevant chapters below. Chapter 5 concerns the nominal subclass nouns, which provide fairly straightforward examples of the morphophonological complexities involved in Pite Saami inflection and derivation, while the nominal subclass pronouns is dealt 4 Morphological patterns and word classes

Nominals I: Nouns
Nouns in Pite Saami form an open class of words that are formally defined by their ability to head a nominal phrase. As the head of an NP, a noun inflects for case and number. Each nouns consists of a lexical stem followed by an inflectional class marker and a portmanteau suffix indicating case and number, as illustrated in (1).
(1) ∑ + class-marker + case/number Pite Saami noun stems can have up to three allomorphic forms throughout the nominal paradigm due to a complex combination of morphophonological processes. The current chapter first describes the morphological categories number ( §5.1) and case ( §5.2) in a general way in order to provide a background for the variables discussed in §5.3 on morphological case and number marking. A description of the inflectional class markers and the resulting inflectional classes for nouns is given in §5.4. 1 The final section ( §5.5) deals briefly with the possessive suffixes, an infrequent set of archaic suffixes that indicate number and case and signify the possessor of the head noun's referent.

Number in nouns
Pite Saami nouns inflect for singular and plural in all grammatical cases except the essive and possibly the abessive case. Dual is not a relevant category for nouns, despite being an integral category in verb morphology and for some pronoun classes. Number is expressed along with case by portmanteau suffixes, stem alternations, or a combination of both. §5.3 on number and case marking treats this in more detail.
There is no formal distinction between count and mass nouns in Pite Saami, as illustrated by the example in (2), in which the words for 'flour', 'sugar' and 'food' are all inflected for plural.

The nominal case system
Pite Saami has nine cases: nominative, genitive, accusative, illative, inessive, elative, comitative, abessive, and essive. 2 Nouns inflect for these cases, in addition to number, via portmanteau suffixes, stem alternations, or a combination of both. A general description of the cases is provided here. Note that the case system is valid for pronouns (also a subclass of nominals) as well, but not for adjectives and numerals. Case is expressed along with number by portmanteau suffixes, stem alternations, or a combination of both. §5.3 on treats this in more detail.

Nominative case
In addition to being used as the citation form, most commonly in singular, nominative case (nom) marks the grammatical subject of a verbal clause (typically the most agent-like argument for transitive verbs) as in (4) and (5) The possessed noun in a possessive copular clause (cf. §13.1.4) is also in the nominative case, as in (6)

Genitive case
The genitive case (gen), the only adnominal case in Pite Saami, marks the possessor modifying the head of a noun phrase (the possessed noun), as in (7)

Accusative case
The accusative case (acc) marks the object of a transitive verb, as illustrated by the monotransitive clause in (9). In ditransitive clauses, the accusative marks the object referring to the theme while the recipient is marked by the illative, as in (10) The accusative can also mark nouns functioning as a clause-level temporal adverbial phrase denoting a period of time, as in (11)

Illative case
The illative case (ill) marks nouns that are the goal of the action expressed by a verb of motion, as in (12) In addition, the illative case marks nouns that refer to the addressee of communication, as in (13), and the recipient of 'giving' actions, as in (14) Finally, familial relations can also be expressed using an illative construction. In such cases, the 'ego' of the family relation is in the illative, as in (15) In this example, the illative nominal is a pronoun, but it is plausible that full nouns are possible in this function as well, although there are no such tokens in the corpus.

Inessive case
The inessive case (iness) marks nouns that function as adjuncts to verbal clauses indicating the location of the event or action, as in (16).

Áhkkabakten?
Ahkkabakte-n Ahkkabakte-iness.sg 'Well, what more did you do in Áhkkabakkte?' [pit080924.021] Similarly, as the complement of the copular verb, an inessive noun indicates the location of the subject referent, as in (17).
The possessor noun in a possessive copular clause (cf. §13.1.4) is also in the inessive case, as in (18)

Elative case
The elative case (elat) marks nouns as the source of an action of transfer, as in (19), as well as the origin, as in (20) and (21) The elative also marks the addressee of a question (the source of information), as in (22).
The noun referring to the material that something consists of or is made of is in the elative case, as in (24)  The elative case can be used to mark the agent which carries out the action referred to by a passivized verb, as in (26) In comparative constructions, elative marks a noun whose referent is the standard in the comparison, as in (27).

Comitative case
The comitative case (com) marks nouns referring to someone or something participating in an action together with the agent as in (28), or some other participant, as in (29) suhkor-ahta sugar-abess 'Do you take your coffee with sugar or without sugar?'

Nominals I: Nouns
The comitative also marks nouns referring to an instrument used to carry out an action, as in (30) When two persons or things are equated with respect to a certain characteristic, the comitative marks the noun whose referent is the standard of comparison, as in (31)

Abessive case
The referent of a noun marked by the abessive case (abess) is lacking or missing, as illustrated by (32)

Essive case
The essive case (ess) generally marks predicative nouns functioning as complements of verbs such as sjaddat 'become', as in (34) and (35), and gähtjoduvvat 'be called', as in (36) (37), her initial response normally consisted of nearly the same construction, only with the noun in nominative case, as in (38) Finally, it should be pointed out that essive is not particularly common in the corpus; tokens for this case are only found in elicitation sessions. In summary,

Nominals I: Nouns
there are not enough data to come to any definitive conclusions concerning the status of essive in current Pite Saami usage.

Number and case marking on nouns
As indicated in the previous sections, Pite Saami nouns inflect for nine cases and two number categories (only the essive and possibly the abessive cases do not inflect for number). While case and number are generally marked by nominal suffixes, they are often supplemented by other morphophonological marking strategies, or even expressed solely by non-linear morphology. These other strategies are: • consonant alternations in the stem (also known as consonant gradation) • stem-vowel alternations (umlaut) • vowel harmony Concerning nouns, the segmental alternations are discussed in detail in §5.3.2, while vowel harmony is presented in §5.4.1.1. First, a short discussion of the nominal suffixes follows here.

Nominal suffixes
Pite Saami has a number of portmanteau suffixes expressing case and number. Only nom.sg, nom.pl and gen.sg are generally not marked by any linear morphology. The nominal suffixes marking case and number are listed in Table 5.1 on the facing page. Note that the status of the abessive suffixes is unclear, including whether they inflect for number, as discussed in §5.3.3. Nouns in essive case do not inflect for number.
In nom.pl and gen.sg, the -h suffix is optional in Pite Saami (and therefore appears in parentheses in Table 5.1). 5 The com.sg suffix has two allomorphs: -jn and -jna, which seem to be in free variation in the corpus, and not determined phonologically.

Nominal suffixes and syncretism
Several of the nominal inflectional suffixes, considered by themselves, are homophonous:  For Class III nouns 7 which do not exhibit any stem allomorphy, the corresponding inflected noun forms within a paradigm are therefore syncretic. Two examples are listed in Table 5.2. However, for nouns which have stem allomorphy (consonant gradation, umlaut and/or j-suffix vowel harmony), different stem allomorphs are chosen for ill.sg than for gen.pl and com.pl, for ess than for iness.sg, and for nom.sg than for nom.pl and gen.sg. 6 The alternative to the optional -h suffix for nom.pl and gen.sg forms is no suffix (except for Class III nouns, which are marked by -Ca). 7 Cf. §5.4.3.

5 Nominals I: Nouns
As a result, only the inflected forms for gen.pl and com.pl, for iness.pl and com.sg (the -jn variant of the latter), as well as for nom.pl and gen.sg are syncretic in all noun paradigms. Some examples are provided in Table 5.3.

Nominal suffixes with a -j component
When looking at the inflectional suffixes, it is noticeable that a number of suffixes contain a -j component, as highlighted by Table 5.4. It is tempting to posit a plural marking suffix -j because it occurs in gen.pl, acc.pl, ill.pl, iness.pl, elat.pl and com.pl; however, the ill.sg suffix -j and the com.sg suffix -jn(a) both have a similar -j element, but are clearly not plural.

Number and case marking on nouns
Furthermore, as illustrated in Table 5.5, the plural cases with a -j component in the suffix trigger vowel harmony in stem consonants in Class Ie nouns (cf. §5.4.1.1), but so does the com.sg suffix, while the ill.sg suffix does not trigger j-suffix Table 5.5: Nominal case/number suffixes with a -j-segment and triggering j-suffix vowel harmony vowel harmony (despite being segmentally identical to gen.pl and com.pl suffixes). Thus, -j suffixes that trigger vowel harmony also fail to align with number marking. As a result, I do not analyze any -j suffix as a plural marker, but do point out this nearly pervasive plural pattern. 8

Non-linear noun morphology
In addition to the suffixes described above, most nouns are also marked for case and number by non-linear stem allomorphy (cf. §4.1.2). Because nom.sg, nom.pl and gen.sg lack suffixes completely, 9 nouns in these three case/number categories can only be marked by non-linear morphology.
To illustrate this, the inflectional paradigm for the noun bärrgo 'meat' is provided in Table 5.6 on the next page and described here. Note that, due to /o/ occurring in V2 position in all forms as the inflectional class marker, the stem has two allomorphs: bärrg-and bierg-.
In summary, the inflectional paradigm for bärrgo is characterized by both consonant gradation and umlaut in the stem, and the morphological environment determines which of these allomorphs is selected. As a result, the acc.pl form biergojd is marked for case/number by the weak bierg-stem and the -jd suffix simultaneously, and the ill.sg form bärrgoj is marked by the strong bärrg-stem and the -j suffix. The most obvious evidence that the choice of stem allomorph is morphologically meaningful can be found in a comparison of the nom.sg form bärrgo and the nom.pl 10 form biergo. These forms differ exclusively in the choice of the strong versus the weak stem allomorph and in the choice of umlaut. Thus, the nom.sg form bärrgo is marked for case/number by the fact that the stem is in the strong grade and features the vowel ä, while the nom.pl stem is in the weak grade and features the vowel ie. This pattern of non-linear case/number marking throughout the paradigm for bärrgo is illustrated in Table 5.7 on the facing page. Here, two patterns are manifest: the forms for nom.sg, ill.sg and ess show one pattern, while all other case/number combinations exhibit the other pattern. This alignment of stem allomorph selection is more or less prevalent throughout Pite Saami noun paradigms whenever stem allomorphy is a part of a noun's inflectional paradigm. Note, however, that not every noun undergoes consonant gradation and/or umlaut; instead, their presence are determined by the phonological form of a noun. Consonant gradation is described in detail in §4.1.2.1, and umlaut in §4.1.2.2; some examples of nouns with consonant gradation and umlaut alternations are shown in Table 5.8 on the next page and Table 5.9 on page 98, respectively.

Problematic case/number marking in abessive case
Unlike the other cases, the behavior of nouns in the abessive case is a bit of an enigma, even if its meaning, which typically translates as 'without', is quite clear (cf. §5.2.8). Indeed, it is difficult to come to any certain conclusions about the rela- Table 5.9: Umlaut alternation patterns for nouns, with nom.sg and nom.pl example pairs x -y nom.sg nom.pl ɛ -e /pɛkːa/ -/pekːa/ 'wind' bägga biegga u͡ a -o /lu͡ akːta/ -/lokta/ 'bay' luakkta luokta tionship between abessive as a case per se and the morphophonological marking of nouns in the abessive case. It seems to be rarely used in natural speech, and is only attested in the corpus in elicitation sessions. Even in elicitation sessions, language consultants were often hesitant or uncertain of the word forms they produced, and often produced conflicting forms for a single item. Indeed, the slipperiness of the abessive case is nothing new, as both Lagercrantz (1926) and Lehtiranta (1992) only provide incomplete treatments of abessive. One potential source of the confusion (even for speakers) is the fact that abessive suffixes are unique in two ways. First, there is significant allomorphy, and, secondly, some of the allomorphs are the only bisyllabic nominal inflection suffixes in Pite Saami. The attested forms are -dak, -daga, -gat, -gahta and -ahta (cf. examples (32) and (33) on page 90). Furthermore, the weak grade usually accompanies abessive, but sometimes the strong grade does. In some cases of Class Ie nouns, j-suffix vowel harmony is triggered, in others it is not. In some cases, number is clearly marked, in other cases, there is no distinction between singular and plural.
As a result, the following sections on inflectional noun classes are only able to provide a limited and preliminary description concerning abessive.

Inflectional classes for nouns
Nouns in Pite Saami can be grouped into three main inflectional classes, with several subclasses, based on recurring patterns across case/number inflectional paradigms. Each noun is marked by a class suffix 11 which is attached directly after the noun stem and precedes case/number suffixes (cf. Figure 1 on page 83).

Inflectional classes for nouns
For the majority of nouns, this suffix consists only of a vowel (in V2 position); however, the class marking suffixes in the less frequent classes II and III deviate from this pattern. The presence of umlaut alternations and/or consonant gradation for a given noun is not dependent on the noun's membership in a specific class, but is determined by whether the phonemes occupying the V1 position and the consonant center of the final foot, respectively, are susceptible to umlaut and/or consonant gradation. Furthermore, some derivational suffixes (such as the diminutive suffix -tj) can block consonant gradation and umlaut from happening in the new derived form. Note that membership in a specific noun class does not seem to be semantically motivated.
The following sections present the four inflectional noun classes based on a preliminary analysis of the corpus; it is possible that, with more research, more noun classes may result, or that the present classes may need revision. Because each noun paradigm consists of seventeen inflectional forms, most of the data on which these classes are based come from elicitation sessions, as it is far beyond realistic for a single, non-native-speaker linguist to collect a sufficiently large natural (i.e., un-elicited, spontaneous) spoken language corpus which includes all inflectional forms for a large variety of nouns.
There are two main criteria for positing the different noun classesː • the allomorphy of the nom.sg form of a noun stem in relation to the rest of the inflectional paradigm (i.e., consonant gradation, umlaut) • the regularity of the pattern of vowels occurring between the stem and case/number suffixes (i.e., the class marking suffix) To illustrate these differences, it is sufficient to look at the class suffix in nom.sg and the alignment of consonant gradation allomorphs, as summarized in Table 5.10. The header grade alignment refers to the choice of stem allomorph in nom.sg versus nom.pl whenever consonant gradation is relevant for a specific

Class I
Nouns in Class I are characterized by: • str-wk grade alignment (when relevant) Class I nouns can be divided into five subclasses, depending on the class-marking vowel they have, as illustrated in Table 5.11.
The first four subclasses behave very similarly, while subclass Ie is unique due to the presence of j-suffix vowel harmony, as discussed in detail in §5.4.1.1. For subclasses Ia, Ib, Ic and Id, the class marking suffix is invariable throughout the paradigm, while the class marking suffix for subclass Ie varies due to j-suffix vowel harmony.
For nouns with consonant gradation and/or umlaut in Class I, nom.sg, ill.sg and ess are in the strong grade and have ä or ua/uä in V1 position, while other case/number slots in a paradigm have the weak grade and ie or uo in V1 position. The gradation pattern and class marking suffixes for Class I are summarized in Table 5.12 on the facing page. Here, V stands for the vowel which comprises the suffix for each class (i.e., a for Class Ia, á for Class Ib, o for Class Ic, å for Class Id and e/i/á for Class Ie).
Inflectional paradigms can be found on pages 101 and 102 to illustrate subclasses Ia, Ib, Ic and Id. The paradigms for the words luakkta 'bay' and mánná 5.4 Inflectional classes for nouns  Table 5.13 as an example for Class Ia and Class Ib nouns, respectively. Table 5.14 shows the paradigms for bäbbmo 'food', a Class Ic noun, and for skåvvlå 'school', a Class Id noun. Because subclass Ie is more complex due to j-suffix vowel harmony, it is discussed separately in §5.4.1.1.

Class Ie
Class Ie nouns are a special subset of class I nouns due to two features: • j-suffix vowel harmony • allomorphy in the class markers (e, i, á) The inflectional paradigm for the words guolle 'fish' and vágge 'valley' are provided as examples for Ie nouns in Table 5.15 on the next page; note that vágge is not subject to consonant gradation. For nouns with consonant gradation and/or umlaut in Class Ie, nom.sg, ill.sg and ess are in the strong grade and have ä or ua/uä in V1 position, while other case/number slots have the weak grade and ie or uo in V1 position, just as with all Class I nouns.
As mentioned above, nouns in Class Ie are subject to j-suffix vowel harmony. This refers to non-adjacent regressive vowel harmony triggered by the presence of /j/ in certain case/number suffixes. In this, certain V1 vowels and the V2 vowel are raised in accommodating the palatal position of the /j/ in the suffix. The vowel in V2, which is e in Class Ie nouns, is raised to i, while the vowel in V1 is raised depending on its initial value: ä→i, uo/uä→u, a→ä, á→ä and å→u. Other V1 vowels in Class Ie are not affected, but the V2 vowel is always raised from /e/ to /i/ in the relevant paradigm slots.
The class marking suffixes and consonant gradation pattern for Class Ie are summarized in Table 5.16 on the facing page.

Inflectional classes for nouns
vágg-e-n

Class II
Two features mark nouns in Class II: • the class marking suffix for nom.sg is -Vj, while for the other case/number slots, the class marker is -V • wk-str grade alignment (when relevant) The inflectional paradigm for the words båtsoj 'reindeer' and ålmaj 'man' are provided as examples in Table 5.17. ålm-a-jn ålm-a-j abess n/a n/a ess ålm-a-n Note that the grade alignment in the båtsoj paradigm is for the most part wk-str, i.e., the weak grade is found in nom.sg and ess, and the strong grade elsewhere. The lexical item ålmaj does not feature consonant gradation, and attempts to elicit the abessive forms resulted in three inconsistent forms. The gradation pattern and class marking suffixes for Class II are summarized in Table 5.18 on the facing page.
There do not appear to be many words in Class II, and the data in the corpus are ultimately inconclusive. There are some irregularities which cannot be explained, for instance why the vowel in V2 position in båtsoj is only o in nom.sg and ess, but otherwise u, while the V2 vowel in ålmaj is consistently a. Furthermore, it is unclear why consonant gradation in the ill.sg form in the båtsoj paradigm does not align with nom.sg and ess.

Class III
Three features mark nouns in Class III: • the stem is consonant-final • the nom.sg form lacks a class suffix • wk-str grade alignment (when relevant) This class consists of two subclasses (IIIa and IIIb), as discussed in the following two sections.

Class IIIa
Class IIIa is the more common Class III subclass. It exhibits a nom.sg form which lacks a class marker and ends in a closed syllable; in this case, the stem-final consonant is thus the word-final consonant. The paradigms for the nouns sabek 12 'ski' and vanás 'boat' are provided in Table 5.19 on the next page as examples for this subclass. The word vanás 'boat' is similar to sabek 'ski', but is subject to consonant gradation. Finally, denominal nouns derived by the diminutive suffixtj are all in Class IIIa. Table 5.20 on the following page provides a nearly complete paradigm for guolátj 'little fish'.
guolátj-i-jn guolátj-i-j abess n/a n/a ess n/a

Class IIIb
The less common subclass of Class III nouns exhibits a nom.sg form which also lacks a class marker but ends in an open syllable; thus, the stem-final consonant, which is present in all other slots in the paradigm, is lacking. The word bena 'dog' is provided in Table 5.21 as a example for this second subclass.
bednag-i-jn bednag-i-j abess n/a n/a ess n/a

Class III summary
The gradation pattern and class marking suffixes for Class III are summarized in Table 5.22 on the following page. As with Class II words, the corpus only provides limited data on Class III words, and attempts to elicit abessive and essive forms led to inconsistent results, partly due to uncertain native speaker intuition for these rare forms. However, elicited abessive forms were consistently in the strong grade, while elicited essive forms were sometimes in the strong grade, sometimes in the weak grade, without any seemingly consistent patterns. 13

Summary of noun classes
str -i-str -iabess str n/a str n/a ess n/a n/a paradigm for each word is not listed due to a lack of space, the forms for nom.sg, nom.pl, acc.sg, gen.pl, ill.sg and elat.sg are more than sufficient to convey the relevant morphological differences between the classes.

Possessive suffixes
A special set of possessive suffixes exists in Pite Saami which indicate, in addition to case and number for the host noun, the person and number of the possessor of the referent of the host noun. While the possessive suffixes go back to Proto-Saami (Sammallahti 1998: 73), they seem to have nearly fallen out of use in contemporary Pite Saami, and are only attested in three recordings from the corpus. These examples from the corpus are presented first, and a discussion follows. While there are technically nine tokens of possessive pronouns in the corpus, these nine tokens can be grouped into two identical sets, so that effectively only two examples are available. Specifically, there are three tokens of áhttjes 'my father' in nominative case by one speaker in two different recordings, and three tokens of the parallel construction mammaset ja pahpaset 'your mother and your father' by another speaker in one recording. An example from the first speaker is provided in (39), and an example 14 from the second speaker in (40).  ja ja and pahpaset pahpa-set father-2pl.poss\ill.sg 'Well then she said "well, say hello to your mother and your father". ' [pit100703a.038] Just as with the other case/number suffixes, the possessive case/number suffixes follow the inflectional class marker, as illustrated in (41). (43) mamm-a-set mother-Ia-2pl.poss\ill.sg 'to your (pl) mother'

Possessive suffixes
(44) pahp-a-set father-Ia-2pl.poss\ill.sg 'to your (pl) father' While these three examples do not provide enough evidence for case and number marking in addition to possession, the thorough paradigm for åbba 'sister' and a very partial paradigm for áhttje 'father' in Lehtiranta (1992: 158-159) 15 indicate that the possessive suffixes are best described as portmanteau suffixes which

Possessive suffixes
indicate the number and case of the host noun as well as the person and number of the external possessor. 16 As substitutes for the external possessor NP, they fill a pronominal function, as well.
It should be pointed out that the possessive suffixes above do not correspond to the equivalent examples in the Lehtiranta paradigms: Lehtiranta has aah'tjaam for 'father-1sg.poss\nom.sg', while aah'tjies is listed as 'father-3sg.poss\nom.sg', a form which is much closer to the form in (42), but means 'his/her father'. 17 Furthermore, Lehtiranta indicates that ååp'paasetieh is 'sister-2pl.poss\ill.sg, which has an additional -ieh word-finally not found in (43) or (44).
In all other cases in the corpus, NP-internal possession is expressed using a noun or pronoun in the genitive case, as in (45), which is from the same speaker and recording as in (40)  It is likely the case that alienability plays (or played) a role in determining which nouns can be marked with possessive suffixes. It is also possible that certain nouns with possessive suffixes have been lexicalized in current usage. While the lack of possessive suffixes in the corpus seems to indicate that they are no longer used regularly, the fact that the two obvious loan words in (43) and (44) 18 have possessive suffixes, indicates that they may still be productive somehow, or at least retrievable via analogy. At any rate, the corpus does not provide nearly enough data on the possessive suffixes and any conclusions on their current state this topic must be left to future research.

Nominals II: Pronouns
Pite Saami has a closed class of pronouns consisting of personal, demonstrative, reflexive, interrogative and relative pronouns. Pronouns are nominals and are defined syntactically by their ability to substitute a nominal phrase. As nominals, all pronouns inflect for case (cf. §5.2 on the case system); concerning number, personal and reflexive pronouns inflect for singular, dual and plural, while demonstrative, interrogative and relative pronouns only inflect for singular and plural. The Pite Saami pronouns are described below, in the order listed above; paradigms for each pronoun type are also included. The pronouns are written using the working Pite Saami orthography. The corpus does not provide sufficient data about the status of any pronouns in the abessive and essive cases, 1 so this must be left for future study. Note that there are also a number of non-nominal interrogative pro-forms which do not inflect for case or number; although not nominals, these pro-forms are covered in §6.4.4, after the interrogative pronouns.

Personal pronouns
Personal pronouns inflect for person and number (singular, dual or plural) as well as for case. They are listed in Table 6.1 on the following page. Personal pronouns do not inflect for the biological sex of their referents, but are restricted in referring only to humans (demonstrative pronouns are used when the referent is not human). The nominative forms all have two possible forms, e.g. mån∼månnå '1sg.nom'. In general, the monosyllabic form is the default, while the bisyllabic form is typically used as a citation form and when the pronoun is emphasized.
The person marking morphemes in personal pronouns are completely systematic and are listed in Table 6.2 on the next page. Case and number marking is not quite as systematic, but certain segmental patterns are present which closely resemble the singular case/number suffixes for nouns, particularly those in inflectional class I (cf. §5.4.1). Specifically, note the lack of a final consonant for all full forms of nominative pronouns (cf. the lack of a nom.sg or nom.pl nominal case suffix), genitive pronouns in dual and plural (cf. the lack of a gen.sg

Demonstrative pronouns
Demonstrative pronouns are based on the stem d-. They inflect for case and number (singular and plural, but not dual), as well as the proximity of the entity they refer to. The data from the corpus indicate that there is a three-way distinction between referents close to the speaker (proximal), those away from the speaker (distal), and those particularly far away (remote). The demonstrative pronouns are listed in Table 6.3. Note that due to a lack of sufficient data on the remote forms in the corpus (either no forms exist or speakers were too uncertain to warrant inclusion here), this part of the paradigm is not complete; forms based on tentative data are marked by a question mark. Morphologically, demonstrative pronouns consist of the stem d-, followed by -á-, -a-or -u-for proximal, distal and remote, respectively. This is then followed by a case/number suffix, as summarized in Table 6.4 on the next page.
Both distal and remote demonstrative pronouns have a referent which is away from the speaker, but remote demonstrative pronouns indicate a greater distance than distal demonstrative pronouns so that the referent of a remote demonstrative pronoun is clearly not located near the addressee. On the other hand, the referent of a proximal demonstrative pronoun is very close to the speaker. Distal demonstrative pronouns are not as specific in indicating the location of the referent, and do not necessarily rule out a referent which is near the addressee. Indeed, distal demonstrative pronouns are the most common in the corpus and are a sort of unmarked default demonstrative pronoun. Note that demonstratives are identical in form to demonstrative pronouns, but differ syntactically because they modify the head of an NP; they are discussed in §7.8. Demonstrative pronouns typically have non-human referents, as in (1).
(1) muhtin muhtin sometimes sa sa so del del then vuoptin vuopti-n sell-1du.pst dajt d-a-jt dem-dist-acc.pl 'So sometimes we sold those. ' [pit080924.300] However, they can also be used to refer to third-person human referents, as in the example in (2).
( Distal demonstrative pronouns can also be used for anaphoric text deixis. For instance, dat in example (3) refers to the fact that the speaker has just dropped her ski pole.

Reflexive pronouns
The reflexive pronouns in Pite Saami are based on the stem etj-and inflect for the number (singular, dual and plural) and person of the noun they are coreferential with. Reflexive pronouns also inflect for case. These are listed in Table 6.5 on the following page. The stem etj-can be translated as 'sel', which could imply that it is a noun, but it is different from nouns for several reasons: 1) it is monosyllabic, 2) it has its own case and number marking suffixes, and 3) it inflects for dual number. Note that reflexive pronouns are not common in the spontaneous language recordings in the corpus, but are mostly found in elicitation sessions. Even in elicitation sessions, my main consultant was not completely sure about some of the forms for less common cases (i.e., everything except nominative, accusative 2 and genitive). Furthermore, a number of the elicited forms deviate from the forms provided in the complete paradigm in Lehtiranta (1992: 162). 3 For these reasons, the forms in Table 6.5 on the next page should be considered preliminary at this point, and potentially subject to modification as a result of more thorough study. My consultants were particularly uncertain about the forms marked by a question mark, while forms listed in parenthesis are not attested in the corpus, but taken from the paradigms in Lehtiranta (1992: 162) and adapted to the current Pite Saami orthography. Reflexive pronouns are frequently used to add emphasis to the noun phrase they are coreferential with (as an intensifier), as in (5)

Interrogative pronouns
The noun phrase that a reflexive pronoun is coreferential with does not have to be realized overtly, as illustrated by the utterance in (6)

Interrogative pronouns
Pite Saami has several classes of interrogative pronouns as well as a set of interrogative pro-forms which do not refer to NPs. While the latter set of non-nominal pro-forms refer to other word classes, they are covered in this section nonetheless due to their syntactic status as pro-forms. The pronouns can be divided into those with human referents (cf. §6.4.1), which use the stem ge-, and those with non-human referents (cf. §6.4.2), which use the stem m-. Furthermore, there are two classes of interrogatives which enquire about the selection of a particular item (semantically equivalent to English 'which'; described in §6.4.3): the first refers to a choice from a selection in general and uses the stem mikkir-, while the other refers to a choice of one or two items and uses the stem gåb-. Interrogative pro-forms not referring to NPs mostly feature the stem g-(cf. §6.4.4). This classification is summarized in Figure 6.1, which also indicates the stem for each type.

Interrogative pronouns with human referents
Interrogative pronouns with human referents use the stem ge-and inflect for the number (singular or plural) of the intended referent and for case. These pronouns 6 Nominals II: Pronouns are listed in Table 6.6; examples are provided in (7) and (8)

Interrogative pronouns with non-human referents
Interrogative pronouns with non-human referents use the stem m-and inflect for the number (singular or plural) of the intended referent and for case. These pronouns are listed in Table 6.7 on the facing page; an example is provided in (9).

Interrogative pronouns concerning a selection
The two selective interrogative pronouns are used to enquire about the selection or choice of an item. The stem mikkir-refers to a selection in general, while the 6.4 Interrogative pronouns stem gåb-limits the selection to one or two choices. These are described in the following two sections.

General selection using mikkir-
Interrogative pronouns based on the stem mikkir-are used to enquire about a choice or selection in general. They inflect for the number (singular and plural) of their referent and for case. These forms are listed in Table 6.8 on the next page. Note that the inessive forms are not attested in the corpus. Two examples are provided in (10) and (11) They can also modify the head of an NP, and are then a 'pro-adjective' 4 enquiring after a further characterization of the referent. In this case, they do not inflect for number or case (as is true of all attributive adjectives), and so the form is always mikkir, as illustrated by the examples in (12) and (13). Another possible form of the stem seems to be makkar-, but this is only attested twice in the corpus and by one speaker, while mikkir-was consistently preferred in elicitation sessions. An example with makkar is provided in (14); here, makkar is a pro-adjective modifying a noun in a subordinate interrogative clause. sarvajd sarva-jd reindeer.bull-acc.pl gilgin gilgi-n will-3pl.pst njuovat njuova-t slaughter-inf aj aj also 'And then he is also checking a bit which reindeer bulls they should also slaughter. ' [pit080909.006] 6.4 Interrogative pronouns

Limited selection using gåb-
A further interrogative pronoun used to limit a selection to only one or two is based on the stem gåb-. It inflects for case and for number (singular and plural), as described below. Table 6.9 lists the various forms. This interrogative pronoun is only attested in elicitation sessions in the corpus. A more thorough description must be left to future research.

Non-nominal interrogative pro-forms
There are a number of non-nominal interrogative pro-forms. These enquire about information typically expressed by a clause-level adverbial, an adjunct or a com-6 Nominals II: Pronouns plement clause. They are listed and glossed in Table 6.10, and three examples are provided in (17) through (19). The list in Table 6.10 is likely not complete, as there are several other nonnominal interrogative pro-forms listed in the Pite Saami wordlist (cf. §1.2.3.4) which are not attested in the corpus. Furthermore, the data do not indicate what the difference is between the various alternate forms for 'to where' and 'how'.

Relative pronouns
Relative pronouns in Pite Saami are identical in form to the interrogative pronouns with non-human referents (cf. §6.4.2). However, unlike interrogative pronouns, relative pronouns do not reflect the human-ness of their referents. They agree in number with their referent, and inflect for the case required by their syntactic function within the relative clause. The relative pronouns are listed in Table 6.11. See §14.2.4 for a number of examples with relative pronouns as well as a description of relative clauses.

Adjectivals
Adjectivals in Pite Saami are defined syntactically by their ability to head an adjectival phrase (AP). They can be divided into four sub-categories based on both syntactic and morphological behavior, as summarized in Table 7.1. While attributive adjectives generally do not inflect, in elliptical phrases in which the head of an NP is not realized overtly, they do inflect for case and number. Predicative adjectives are marked for number, and are morphologically similar to nominals. Demonstratives agree in number and case with the noun they modify. Numerals, on the other hand, are consistently uninflected. Finally, the two types of adjectives form an open sub-class, while numerals are a closed sub-class.
The rest of this chapter covers adjectives, demonstratives and numerals as follows: §7.1 provides a description of attributive adjectives, while §7.2 deals with predicative adjectives, before §7.3 takes up the formal relationship between these two types. §7.4 then goes on to describe comparative and superlative forms, before §7.5 illustrates the implementation of such forms in making comparisons. Syntactic restrictions on the adjectives corresponding to 'small, little' are described in §7.6. Quantifiers (a semantic sub-class of adjectives) are discussed in §7.7, while demonstratives are presented in §7.8. Finally, §7.9 covers numerals.

Attributive adjectives
Attributive adjectives form the head of an AP modifying the head of the matrix nominal phrase, and are normally not subject to inflectional morphology. As part of an attributive AP, an attributive adjective occurs before the head noun it modifies, but after a demonstrative, if present (cf. §11.2 on the structure of NPs). Examples are provided in (1)  As the head of an AP, attributive adjectives can be modified by adverbs of grade, as illustrated by the AP hoj buorak 'really good' in (4), or nåv njalga 'so tasty' in (2)  A number of adjectives end in -s (cf. the two adjectives tjähppes 'black' and villges 'white' in (1)), which is often considered an 'attribution' marking suffix in the literature. 1 However, as njalga 'tasty' in (2) illustrates, not all adjectives 7.1 Attributive adjectives are marked this way. Furthermore, corresponding predicative adjective forms (cf. §7.2 below) often also have a final -s, sometimes even to the exclusion of the attributive adjective form. Because no consistent relationship between forms with and forms without a final -s exists, it is no longer a productive way to mark or derive either attributive or predicative adjectives. For this reason, it is not considered to be morphologically meaningful in the present discussion. Nonetheless, it is worth noting that adjectival forms (both attributive and predicative) ending in -s are common.
Note also that some attributive adjectives appear to have two possible forms. For instance, guhka and guhkes 'long' were both encountered in elicitation sessions with a single speaker who insisted that both forms were equally valid (cf. recording pit080819a starting at 33m14s).

Attributive adjectives in elliptic constructions
If the context of the wider discourse is sufficiently unambiguous, it is possible that the nominal head of an NP is not realized, but implied. When such elliptic constructions (sometimes called 'headless NPs') feature an AP, the head adjective is then inflected for case and number.
For instance, in the elliptic construction in (5), the adjective ruopsis 'red' is marked for singular number and as the object of the verb form bårov 'I eat' for accusative case. In (6), the subject NP also lacks an overt nominal head, but features the adjective tjähppis 'black' (cf. the example in (1) above), which receives nom.pl marking. Adjectives in elliptic NPs can be preceded by a demonstrative, as in (7).
"the system of attributive and predicative marking is highly irregular in the Saamic languages" (Rießler 2011: 215 As the host of case/number inflection, such adjectives look morphologically like nouns. However, syntactically, these adjectives remain adjectives for two reasons. First, they can be modified by adverbs of grade, while nouns cannot be. Second, they generally have a referential antecedent that is the bearer of the property they denote. 2 Semantically, they do not denote an entity (as nouns do), but a property, as with all adjectives. A further example is provided in (8). Here, the choice of the attributive adjective form tjábba 'beautiful' (as opposed to the corresponding predicate adjective form tjábbe) indicates that this is indeed an elliptical NP construction, and not predication.

Predicative adjectives
While attributive adjectives form the head of an attributive AP embedded in an NP, predicative adjectives form the head of an AP which is the complement of the copular verb årrot 'be' and ascribe a property to the subject referent. In (9) and (10), for instance, the predicative adjective corresponding to the attributive adjective tjähppis 'black' (cf. the example in (1) above) is tjáhpat. Morphologically, predicative adjectives are much like nouns because they inflect for number. In fact, many predicative adjectives inflect for number in ways that clearly align with the nom.sg∼nom.pl inflectional marking of certain noun classes. The case could be made that they also inflect for case, although they are always in nominative case (cf. §13.1.4 on copular clauses). However, because no paradigmatic opposition to other case forms exists for predicative adjectives (they are only attested in the corpus as a nominative complement to a copular clause), I conclude that they only inflect for number.
Nonetheless, these are syntactically adjectives, as they head adjectival phrases and can be modified by adverbs of grade, such as nav 'so' as in (11) Table 7.2 on the next page lists a number of attributive adjectives and the corresponding predicative adjectives; the latter clearly align with noun classes in their number marking. The table is divided into subgroups of word forms (indicated by small Roman numerals) that feature the same morphological relationship between attributive and predicative adjectives.
As is evident from the examples in Table 7.2, the attributive forms and the predicative forms correspond in a variety of ways. These correspondence patterns (numbered i-vii) are described here:  iii The attributive form and the singular predicative form are syncretic and have a stem-final -s, while the plural predicative form's stem is also identical, but marked for plural by a final -a. Number marking like class IIIa nouns. iv The attributive form and the singular predicative form are syncretic and have a stem-final -s, while the plural predicative form is marked by a final -a instead of the final es in the other forms. Only one example in the corpus; number marking like a class II noun.

Predicative adjectives
v The bare stem in the singular predicative form, an additional final -is for the attributive form and -a for the plural predicative form. Number marking like class IIIa nouns. vi The singular and plural predicative forms differ only in the choice of stem allomorph (in the consonant center), while the attributive form is in the 'weak' grade (like predicative plural) but with a stem-final gis.
Only one example in the corpus; number marking like a class Ie noun. vii The predicative forms have a stem-final -t or -s, which is lacking in the attributive form. The plural predicative form is marked with a final -a.
In the case of galbma, the attributive form and plural predicative form have the 'strong' stem form, as opposed to the singular predicative form, which is 'weak'. Number marking like class IIIa nouns.
Despite the similarities with nouns described above, it is important to point out that there are a number of predicative adjectives which do not inflect for number. Moreover, this lack of number marking cannot be assigned to any specific noun class, particularly since noun classes with similar segmental structures exhibit clear number marking strategies. Examples of such predicative adjectives are presented in Table 7.5 on the following page.
The paradigms in Table 7.5 are divided into two sub-groupings, again based on the relationship between the attributive form and the predicative forms. They are summarized here: The variety evident in morphological correlations between the attributive and the predicative forms indicates that there is no regular form-to-function relationship between the attributive and predicative forms. Therefore, on formal grounds, the attributive and predicative forms of these property words are assigned to different, though formally and semantically related, adjectival lexemes, as argued for in the following section ( §7.3).

A note on attributive and predicative adjectives
In the literature on Saami languages, a convention prevails by which predicative adjectives are treated as having more or less derivable attributive forms. 4 From a historical point of view, this may be reasonable, particularly if there was a point in the history of the Saami languages at which attributive forms were derived by adding -s and selecting the phonologically relevant stem allomorph, and thus the attributive forms were derivable from the predicative forms. For Pite Saami, however, there is no clear or consistent morphological relationship synchronically between attributive adjectives and the corresponding predicative adjectives, as shown above. This is particularly exemplified by the existence of more than one acceptable attributive form, (as pointed out in §7.1 for the attributive adjective forms guhkes∼guhka 'long'), as well as by the existence of a number of predicative forms ending in -s, but attributive forms lacking -s (cf. pattern vii above). Due to cases like those illustrated by subgrouping ix in Table 7.5, it is not clear that it is sensible to claim that all adjectives have corresponding predicative adjectives that differ at all. Because of the wide variety of and the inconsistencies in morphological patterns between corresponding attributive and predicative adjectives, it is ultimately more elegant to analyze these two sets of adjectives simply as semantically and etymologically related -but not morphologically derivable -adjectives.

Comparatives and superlatives
The comparative and superlative forms of attributive and predicative adjectives are derived using suffixes. It seems that, morphosyntactically speaking, comparative and superlative forms can be derived from all adjectives, even when a semantic restriction could lexically prevent such forms from occurring; cf., e.g., guäktegierdakap 'more pregnant' (pit090927.07m01s). The singular comparative or superlative predicative form is identical to the respective comparative or superlative attributive form. However, the plural comparative or superlative predicative form is always marked by a suffix consisting of a single vowel (mostly -a). In many cases, the stem to which comparative and superlative suffixes are attached is identical to the stem of the positive plural predicative form, but a number of exceptions exist. Table 7.6 on the next page provides some example paradigms. To help illustrate the morphophonemic relationship to positive forms, the singular predicative adjective form is also indicated. Furthermore, the paradigms are divided into subgroupings (each marked with a Roman numeral) based on suffix allomorph patterns. The third and fifth columns in Table 7.6 provide the singular comparative and superlative adjectives (the attributive and predicative singular forms are syncretic), respectively, while the fourth and sixth columns only indicate the suffix used to mark the plural predicative comparative and superlative adjectives, respectively. Note that there are allomorphic alternations in the superlative suffix for subgroupings iii and iv.
Comparative adjectives are derived in a relatively straightforward way: the suffix -p 5 is added to an adjective root. If the root has a closed final syllable, then an epenthetic vowel -u-is inserted between the root and the suffix. In predicative position, plural is always marked by a suffix consisting of a vowel; in most cases (groups i, ii and iv), the vowel is -a, but sometime it is -o (group iii). It is not clear what determines the choice of plural suffix for comparative forms. While all forms marked by -o in the corpus have a stem final -s, not all forms with a stem final -s are marked by -o (cf. nanos 'strong').
The superlative suffix has four allomorphs. For the attributive and the singular predicative forms, the allomorph -mos is chosen when the root has a closed final syllable, as in groups iii and iv. Roots with an open final syllable have either the superlative suffix allomorph -mus or -jmus; however, it is not clear what drives the selection of these latter two allomorphs. The allomorph -bmus-occurs whenever the resulting form has an odd number of syllables, as is the case for roots with a final odd syllable in the plural predicative form. Essentially, the superlative suffix always forms the final foot of a word, and thus is the location for consonant gradation alternations. If a final, odd syllable is present (e.g., for the plural predicative form), then the -bmus-allomorph is chosen. 6 Examples for comparative and superlative adjectives in attributive position can be found in (12) and (13), respectively. Instances for predicative usage can be found in (18) and (19)

Comparing NP referents
As with positive adjectives, comparative and superlative adjectives can occur in elliptic NPs, in which case they inflect for case and number (cf. §7.1.1). Examples are provided in (14) and (15). Note that in the second example, the superlative suffix allomorph is -bmus because the adjective has an odd number of syllables.

Comparing NP referents
Predicative adjectives can be used to compare the referents of nominal phrases. If both referents are considered equal concerning the characteristic of comparison, then the NP of comparison is the subject of a copular predicate which is complemented by a construction using the numeral akta 'one' and the relevant predicative adjective, while the NP of reference is in the comitative case. An example is provided in (16). In such constructions, akta can be shortened to akt. When comparing two referents that are not considered equal, the NP of comparison is the subject of a copular predicate which is complemented by a comparative predicative adjective and the NP of reference in the elative case, as in the example in (18). To indicate that a referent is the most extreme concerning the characteristic of comparison (at least within the group being compared), the relevant NP is the subject of a copular predicate which is complemented by the superlative predicative adjective. The quantifier gajk 'all' can be added for emphasis, 7 as in (19)

Restrictions on smáva and unna 'small'
The paradigms for the two Pite Saami words for 'small' provided in Table 7.5 on page 134 do not sufficiently indicate the restrictions placed on these specific adjectives. The root smáv-, a North Germanic loan word, only modifies plural nouns, while unn-, the native word (Sammallahti 1998: 265), usually only modifies singular nouns. No other Pite Saami adjectives underlie such a restriction; however, the Swedish adjective små (cognate with the source of smáv-) is also restricted to modifying plural nouns. 8 Therefore, it seems that this syntactic restriction was probably also borrowed. While a few examples exist in the corpus of unn-modifying a plural noun, smáv-is the preferred item and much more frequent in the corpus. 9 Examples are provided in (20) through (23). 7 The construction gajk vuorasumos 'absolute oldest' in (19) is possibly a calque based on North Germanic; cf. Swedish allra äldst 'absolute oldest'. In both cases, the adverbial modifier is based on the word for 'all' and precedes the superlative adjective. 8 The Swedish adjective stem lite-is used for singular nouns. 9 A corpus search (including elicitation sessions) resulted in 1 token of unn-and 10 tokens of smáv-modifying a plural noun (carried out on 12 th November 2012). In non-elicited tokens from the corpus, nouns modified by the adjective unna are always diminutive nouns, as in (20) above and in (24) and (25)

Quantifiers
However, when a quantifier is in an elliptic NP, it inflects for case and number (just as with other attributive adjectives). This is illustrated by enabu 'more' in (32), by gajk 'all' in (33), and by nagin 'some' in (34)  The quantifier aktak 'none, any' is used to emphasize a negated clause. It seems to be composed of the numeral akta 'one' and the suffix -k, which is a nominalizer in other cases; however, as illustrated by the examples in (35) and (36), it heads an attributive AP and does not inflect for case and number, unless it is in an elliptic NP, as in (37)

Adjectivals
Concerning the status of corresponding predicative quantifiers, there is not enough data in the corpus to come to a certain conclusion. However, at least the attribute adjective edna 'many, much' corresponds to the predicative adjective form ednak; this is illustrated by (38) and (39). This indicates that attributive and predicative forms of quantifiers also differ in form, just as with other attributive and predicative adjective sets.

Demonstratives
Demonstratives modify a noun phrase by further specifying the head noun concerning the distance of the referent relative to the speaker. Just as with demonstrative pronouns, the corpus data indicate that there is a three-way distinction between referents close to the speaker (proximal), those away from the speaker (distal), and those particularly far away (remote). Indeed, they are identical in form with the demonstrative pronouns listed in Table 6.3 on page 115 in the section on demonstrative pronouns ( §6.2), and are therefore not listed separately here. Unlike adjectives, demonstratives always agree with the noun they modify in number (singular and plural, but not dual) and in case. Examples of demonstratives are provided in (40)

Numerals
Numerals in Pite Saami form a closed class and a distinct closed sub-class of adjectives. Syntactically, they are adjectives because they head an adjectival phrase; however, morphologically, they differ from other adjectives by never inflecting (neither for number in predicative APs, nor for case and number in ellipsis constructions). Furthermore, numerals do not consist of attributive/predicative sets differing in form. Instead, numerals are consistent in form, regardless of being in attributive or predicative position.
Pite Saami numerals form a decimal system consisting of the basic numerals for the numbers one through ten, hundred and thousand. All other numerals are compounds based on these basic terms, with the exception of nolla 'zero'. Basic and complex numerals are dealt with in §7.9.1 and §7.9.2, respectively; the derivation of ordinal numerals is described in §7.9.1.1.

Basic numerals
The basic numerals for the numbers one through ten in Pite Saami are reconstructable native Saamic numerals, and tjuohte 'hundred' is at least from Proto-Saami. 10 The numerals nolla 'zero' and tuvsan 'thousand' are likely more recent borrowings, although it is not entirely clear whether they are from North Germanic or Finnic. 11 These basic cardinal numerals are listed on the left side of Table 7.7 on the following page. 10 Sammallahti (1998: 234-235) indicates that tjuohte 'hundred' was originally a borrowing from Proto-Indo-European into Proto-Finno-Ugric or Proto-Finno-Saamic. 11 The entries for the numerals 'zero ' and 'thousand' in Álgu (2006) only provide etymologies for North Saami and Inari Saami; however, while Finnic is clearly a contact language for these two languages, North Germanic is a contact language for Pite Saami, and therefore also a potential source for these two numerals; cf. Swedish nolla 'zero' and tusen 'thousand'.

Derivation and suppletion in ordinal numerals
In general, ordinal numerals, which are listed on the right side of Table 7.7, can be derived from the corresponding cardinal numeral by replacing the vowel in V2 position and any final consonant with the suffix -át (and its allomorph -et in giehtjet 'seventh'). In addition, the weak stem allomorph is selected and umlaut of V1 occurs, if applicable (cf. §4.1.2 on stem allomorphy). The ordinal numerals corresponding to tjuohte 'hundred' and tuvsan 'thousand' are not attested in the corpus. However, there are exceptions. First, the ordinals vuostas 'first' and mubbe 'second' are suppletive forms compared to the corresponding cardinal numerals akkta 'one' and guäkte 'two'. 12 These two ordinals are used exclusively for the single-digit numbers 'first' and 'second'; any ordinal numeral referring to a number of two or more digits uses a form derived from the cardinal numeral, as described above. This is illustrated in Table 7.8 on the facing page.
Second, the cardinal numeral giehtjav 'seven' differs in the final two segments from the ordinal numeral gietjet 'seventh' (i.e., -av and -et).

Complex numerals
Any numerals other than those listed in Table 7.7 on the preceding page are complex numerals formed by combining the basic numerals. Multiples of ten are composed of the relevant cardinal numeral followed by lågev 'ten'; examples are provided in Figure 7.1.

guäkte-lågev gålbmå-lågev nällje-lågev etc.
two-ten three-ten four-ten 20 30 40 Figure 7.1: Multiples of ten Note that lågev is often shortened to låk in fast speech, as in (43) There are two ways to compose two-digit numerals that are not multiples of ten. One method appends the relevant numeral representing the 'ones-digit' to the multiple of ten, while lågev 'ten' is shortened to låk. This is illustrated in Figure 7.2 on the next page, with examples of two-digit numerals from the corpus presented in (45)  Alternatively, complex numerals may be formed phrasally. According to this strategy, the 'ones-digit' precedes a postpositional phrase headed by the postposition nanne 13 'on' with the multiple of ten as the dependent låge (in gen.sg case), as illustrated in Figure 7.3. However, this latter method was only attested in elicitation sessions with one consultant, and is not found in non-elicited data from the corpus. Native ordinal numerals referring to numbers between ten and one hundred are only attested in the corpus in elicitation sessions, and speakers are quite inconsistent and unsure about them. The same is true for cardinal numerals larger than one hundred. The only example in the corpus for a numeral larger than one thousand is not native, but a Swedish borrowing (in an NP with Pite Saami case and number marking); this is provided in (46). 14 13 Note that nanne is often shortened to nan in rapid speech. 14 With the exception of the case/number suffix -n, the entire phrase nittonhundratalan 'in the nineteen-hundreds' in (46) is borrowed from Swedish nittonhundratalet 'the nineteenhundreds'.

Numerals and morphosyntax
Numerals are generally not subject to inflectional morphology. This is illustrated by the examples (43) through (44) above as well as in (47)

Adjectivals
However, there are at least two exceptions. First, the numeral akta 'one' inflects for acc.sg case when modifying a noun, as illustrated by (51), as well as when it is in a headless elliptical construction, as in (52) 16 With this in mind, the word akta forms a word-class of its own, strictly speaking.

Verbs
Verbs in Pite Saami form an open class of words which are defined syntactically by their ability to head a verb complex, as well as morphologically by inflecting for person, number, tense and mood. Verbs consist of a stem which is followed by a class marker and an inflectional suffix or suffixes, as illustrated in (1). (1) ∑ + class-marker + mood/tense/person/number Verb stems can have up to five allomorphic forms throughout the verbal paradigm due to a complex combination of morphophonological alternations. Verbs form at least five inflectional classes. The inflectional suffixes are exponents for person, number, tense and/or mood. Pite Saami distinguishes three number categories (singular, dual and plural), two tense categories (present and past) and the three modal categories (indicative, imperative and potential).
The first sections of this chapter ( §8.1 on the inflectional categories number, tense and mood; §8.2 on non-finite forms and periphrastically marked categories of future, aspect and negation; §8.3 on passive voice) provide a description of relevant morphological categories as a background for the discussion of morphological marking strategies for verbs in §8.4. Finally, §8.5 draws on the initial sections to posit inflectional classes for verbs.

Person and number
All finite verbs agree in number with the subject of the clause and inflect for singular, dual or plural. Finite verbs in the indicative and the potential mood also agree in person. Inflectional morphology is present even if the subject of the clause is not overt. For instance, in (2), the finite verbs minne and gillge both agree with da, the 3pl subject; in (3), the finite verb lijmen agrees with the 1du subject månnå ja Jåssjå.
(2) ja ja and Note that there are a few examples in the corpus in which speakers do not consistently inflect for dual, but instead use the corresponding plural form.
The imperative is not marked for person, but distinguishes the three number categories singular, dual and plural. For example, in (4) the finite verb tjaske is inflected for the implied (2 nd person) singular subject.

Tense
For indicative clauses, verbs can inflect for present tense, as in (2), or past, as in (3) above. Verbs marked for present tense generally signify that a situation is true in the present, as in (5) below, or they express general truths, as in example (2) above (which indicates a general truth about wolves' behavior). However, present tense can also be used to indicate historical present, as in (6), or planned future situations, as in (7). It is therefore not strictly a present tense and could be considered non-past. Nonetheless, the glossing standard 'prs' is chosen to mark this, as it covers the most common function.

Mood
Pite Saami has three moods: indicative, imperative and potential. Indicative mood is by far the most common mood and is considered the default, unmarked mood, as it is not overtly expressed morphologically, as in the examples in §8.1.2 above. The following two sections deal with imperative and potential mood.

Imperative mood
Verbs inflectedisinflection!verbal for imperative mood indicate that the speaker is instructing or commanding the addressee to carry out the action referred to by the verb; the implied subject is always 2 nd person. Verbs in the imperative are not marked for person, but do inflect for number (singular, dual and plural), as 8 Verbs in (4) above as well as in (8) and (9)  The example in (10) below indicates that imperative can also be used as a kind performative speech-act. Note that Lehtiranta (1992: 150-155) includes a second imperative category in his verb paradigms that inflects for all three person categories and is marked by a stem-final -u-; Lehtiranta terms this 'imperative II'. Lagercrantz (1926: 22) mentions 'imperative II' in passing as well, explaining that it is "less severe and more like a wish" (my translation), but Lagercrantz only includes examples for 2sg. The Pite Saami Documentation Project corpus does not have any tokens of such verbs, so more study is needed to determine their current status.

Potential mood
Verbs can also be inflected for potential mood, indicating that the action referred to by the verb is likely to happen. Verbs in the potential mood are marked by a linearly segmentable morpheme -tj-followed by a person/number suffix. 2 Examples are provided in (11) through (13). As the examples in (14) and (15)  The person/number suffixesisinflection!verbal for potential mood are homophonous with those used in present tense for Class V verbs (cf. §8.5.5); cf. §8.4.3 for a discussion of the status of verbs in the potential mood as inflectional and derivational forms.

Non-finite verb forms and periphrastically marked verbal categories
A number of non-finite verb forms exist in Pite Saami. The most common of these are the infinitive, connegative, perfect and progressive forms. Each of these nonfinite verb forms can co-occur with an auxiliary verb to periphrastically express the verbal categories of future tense, perfect or progressive aspect, and negation;  suffix -m, strong grade co-occurs with auxiliary årrot 'be' progressive suffix -min, strong grade co-occurs with auxiliary årrot 'be' The literature on Saami languages often treats non-finite verb forms in addition to those mentioned above. These include the verb genitive, verb abessive or gerunds, for instance. 3 For Pite Saami, Lehtiranta (1992: 95-106) describes the morphological form a number of such non-finite forms, 4 while Lagercrantz (1926) does not describe such verb forms.
With this in mind, it is certainly plausible that Pite Saami has other non-finite

Non-finite verb forms and periphrastically marked verbal categories
verb forms other than those mentioned here. However, there is no evidence of such forms in the present corpus. Ultimately, the morphological and syntactic behavior of other non-finite verb forms must be left for future study.

Future
The verb gallgat 'will' plus the infinitive form of the lexical verb can together express a future activity. The examples in (16)  årro-t be-inf 'How many days are you going to be there?' [pit080924.658] Note that, as mentioned in §8.1.2 above, the present tense is also used to express planned future events.

Aspect
Pite Saami features two aspects, perfect and progressive, as described in §8.2.2.1 and §8.2.2.2 below. Both aspects are formed periphrastically using a combination of the auxiliary verb årrot 'be' and the relevant non-finite verb form. See also §13.1.5.2 on the syntactic structure of clauses with perfective and progressive verbs.

Perfect
The perfect verb form is marked by the suffix -m (glossed as prf); the verb stem is in the strong grade when consonant gradation is relevant. Verbs in the perfect generally indicate that an action in the past still has relevancy in the present situation. For instance, in (19) the speaker is slaughtering a reindeer, and is now able

Progressive
Verbs in the progressive indicate that an activity is ongoing. The progressive verb form is marked by the suffix -min (glossed as prog) appended to the verb stem, which is in the strong grade when consonant gradation is relevant. In (19) above, the speaker uses the progressive form rhtastemin because he is in the middle of 8.2 Non-finite verb forms and periphrastically marked verbal categories cutting out the stomach as he utters the sentence. In (21), the speaker's mother is still living, as opposed to the deceased. The action expressed by a progressive verb does not have to be simultaneous with the moment of the utterance, but can be past tense, as shown by the example in (22). Here, the speaker is describing a picture which was taken while picking blueberries.

Progressive verb forms used adverbially
The progressive form of a verb can also be used in an adverbial function. For instance, tjájbmamin 'laughing' in (23) and gullamin 'listening' in (24) are each used as a modal adverbial to indicate a simultaneous activity.

Negation
Negation in Pite Saami is expressed periphrastically by a finite negation verb and a non-finite verb form. The inflectionalisinflection!verbal behavior of the negation verb is presented in §8.5.8, while syntactic aspects of negation in Pite Saami are covered in more detail in §13.1.5.3; however, a brief description of negation is provided here.
As with any finite verb, the negation verb agrees in person and number with the subject of the sentence and inflects for tense or mood. The complement verb occurs in a special non-finite verb form called the connegative (glossed as 8 Verbs conneg), which is in the weak grade (when gradation is relevant) and otherwise lacks any additional morphological marking. Examples for present and past indicative as well as imperative forms are provided in (25)

Passive voice
Verbs in the passive voice can be derived from other verbs by the derivational suffix -duvv. Note that the vowel immediately following this suffix is the class marking morpheme for Class IV verbs; cf. §8.5.4. Examples are provided in (28)  The data from the corpus concerning passive verbs are quite limited, but indicates that passive verbs can be finite verbs inflecting for tense, person and number, as in (28), or non-finite forms, such as the perfect, as in (29) and (30). However, due to a lack of data, it is not clear whether passives can be used for progressive aspect, or inflect for either imperative or potential mood.
That being said, these examples do make clear that the passive marker is restricted to lexical verbs. Passives are therefore not considered to be part of inflectional paradigms, but instead valency-decreasing verbal derivations. See also §10.2.5 in the chapter on derivational morphology and §13.1.1.1 on syntactic aspects of clauses in the passive voice.
Note that Ruong (1945) includes other derivational suffixes which create passive verbs that are not attested in the corpus.

Morphological marking strategies on verbs
As shown in §8.1 above, finite verbs can be marked for four inflectional categories: • agreement in person with the subject • agreement in number with the subject • tense • mood

Verbs
Just as with nouns, inflectional categories for verbs can be expressed by suffixes and by non-linear morphology, and frequently a combination of both. In the following, §8.4.1 focusses on inflectional suffixes, while §8.4.2 goes on to describe the behavior of non-linear morphology found in stem-consonant alternations (consonant gradation), stem-vowel alternations (umlaut), and vowel harmony. The final section (8.5) then uses the various morphophonological inflectional patterns found across verb paradigms to posit five preliminary inflectional classes for verbs.

Inflectional suffixes for verbs
The portmanteau suffixes expressing agreement in person and number as well as tense or mood in finite verbs are listed in Table 8.3. In this table, if only one suffix is given in a slot, then it is found in all inflectional classes. When more than one suffix is included in a slot, then the first allomorph is for inflectional classes I, II and III, the second allomorph for class IV, and the third allomorph for class V verbs. The suffixes for the non-finite infinitive, connegative and perfect verb forms are included here and in the following sections because they are common verb forms in the corpus and particularly useful in recognizing patterns in verb paradigms. -v -n/-jin/-n -p 2 nd --bähten/-bähten/-hpen -ähtet/-bähtet/-hpit 3 rd -/-ja/--ba -/-je/pst 1 st -v/-jiv/-jiv -jmen -jmä/-jme/-jme 2 nd -/-je/-je -jden -jdä/-jde/-jde 3 rd -j -jga -n/-jin/-n imp 2 nd --n/n/a/-hten -t/n/a/-htet inf -t conneg prf -m

Verbal suffixes and syncretism
Several of the verbal inflectional suffixes, considered by themselves, are homophonous:

Morphological marking strategies on verbs
• -v for 1sg.prs and 1sg.pst in classes I, II and III • -n for 1du.prs, 3pl.pst in all classes, and also du.imp in classes I, II and III • -t for inf and pl.imp in classes I, II and III • no suffix for 2sg.prs, sg.imp and conneg in all classes; 3sg.prs, 3pl.prs in classes I, II, III and V; and 2sg.pst in classes I, II and III Despite these similarities, only the morphology of 1du.prs and 3pl.pst verb forms is syncretic in all verb classes because in most cases homophonous suffixes combine with different non-linear morphology and/or with different class marking suffixes.

Non-linear morphology in verbs
In addition to using the inflectional suffixes described above, inflectional categories for verbs can be marked by one or more of the following stem allomorphy strategies: • stem consonant alternations (consonant gradation) • V1 vowel alternations (umlaut) • V1 vowel raising when followed by a close/close-mid V2 vowel (vowel harmony) Because 2sg.prs, 3sg.prs, 3pl.prs, 2sg.pst, 3sg.pst, sg.imp and conneg forms often lack suffixes (cf. §8.4.1.1 above), verbs in these inflectional categories are typically marked exclusively by these essentially non-linear morphological marking strategies. To illustrate this, the inflectional paradigm for the verb buälldet 'ignite, burn' is provided in Table 8.4 on the next page and described here. Note that the vowel in V2 position (a, e and i) in all forms is the inflectional class marker for Class III verbs (cf. §8.5.3); thus the stem has five allomorphs: buälld-, bualld-, buold-, bulld-and buld-. 5 This reflects a consonant gradation pattern that alternates between strong lld and weak ld, and an umlaut pattern that alternates between ua/uä and uo in the vowel in V1 position. 6 Furthermore, the forms for 1du.prs, 3pl.prs and all past forms are subject to vowel harmony; 8 Verbs buold-a-v bulld-e-n buälld-e-p 2 nd buold-a buälld-e-bähten buälld-e-bähtet 3 rd bualld-a buälld-e-ba bulld-e pst 1 st bulld-i-v buld-i-jmen buld-i-jmä 2 nd bulld-e buld-i-jden buld-i-jdä 3 rd buld-i-j buld-i-jga bulld-e-n imp 2 nd buold-e buälld-e-n n/a inf buälld-e-t conneg buold-e prf bualld-a-m here, the vowel in V1 position is raised to u in the presence of a close-mid front (e) or a close front (i) vowel in V2 position. Note, however, that this vowel harmony is morphologically selected by these slots in the paradigms; the e in V2 in other inflected forms does not trigger vowel harmony (cf. 2du.prs or du.imp forms).
In summary, the inflectional paradigm for buälldet 'ignite, burn' is characterized by consonant gradation, umlaut and vowel harmony in the stem, and the morphological environment determines which of these allomorphs is selected. For instance, as a result, the 1sg.prs form buoldav is marked for person, number and tense/mood by the weak buold-stem (with the -uo-umlaut form) and the -v suffix simultaneously, and the 1pl.pst form buldijmä is marked by the weak buldstem subjected to vowel harmony, and the -jmä suffix.
The pattern of non-linear inflectional marking throughout the paradigm for buälldet is illustrated in Table 8.5 on the facing page. The patterns for both consonant gradation and for umlaut in verb classes subject to these morphophonological strategies align seamlessly. However, each of the two verbal inflection classes subject to vowel harmony has its own unique vowel harmony pattern. Not every verb undergoes consonant gradation and/or umlaut; instead, their presence are determined by the phonological form of a verb. 7 Some examples of verbs with umlaut alternations and consonant gradation are shown in Table 8.6 below and Table 8.7 on the next page, respectively. Note that ua and uä are allophones of /ua/; cf. §3.2.1.9. Table 8.6: Umlaut alternation patterns for verbs, with 3sg.prs and 2sg.prs example pairs

Vowel harmony patterns for verbs
Vowel harmony in verb forms refers to a regressive assimilation of place of articulation between the two vowels of the final foot in a word. Specifically the raising of the vowel in V1 position triggered by the presence in specific, classdependent paradigmatic slots of a close-mid /e/ (orthographic e) or a close front /i/ (orthographic i) vowel in V2 position. There are six attested vowel harmony patterns in the V1 vowel of a verb stem from Class II or Class III, as illustrated by Table 8.8 on the facing page. The data from the corpus indicate that Class I and Class IV verbs do not exhibit vowel harmony, but there are no tokens of Class I or Class IV verbs with one of the vowels listed in Table 8.8 on the next page in V1 position. Consequently, the data must be considered inconclusive in this respect. On the other hand, it is quite evident that Class V verbs are not affected by vowel harmony because the V2 vowel in Class V verbs is never subject to the allomorphic alternations which trigger vowel harmony in the V1 vowel.
It is not clear why á and a have different vowel harmony alternations (i/ä and i/e, respectively, as illustrated by the first four examples in Table 8.8); these alternation patterns do not align with verb classes. Further research is needed to come to a better understanding of this vowel harmony.

The potential mood: inflection or derivation?
The potential mood 8 is not attested very often in the corpus, particularly outside elicitation settings, and was not considered in most elicitation sessions focussing on verb paradigms. As a result, the amount of data from the corpus available to inform a description of the inflectional behavior of the potential forms are quite limited. Nonetheless, the paradigm for the potential forms of the verb gullat 'hear' is provided in Table 8.9. Taking the potential forms presented in the verb paradigms in Lehtiranta (1992: 150-155) and in the examples in Lagercrantz (1926: 22-24) into consideration, the paradigm of class marking suffixes and person/number suffixes used for potential verb forms is presented in Table 8.10 on the next page. The stem allomorph of the verb is in the weak stage, when applicable.
In the literature on Saami languages, potential mood is normally treated as an inflectional category, 9 and, for this reason as well as due to its seeming opposition 8 Verbs Table 8.10: Class marking suffixes and person/number suffixes for potential verb forms sg du pl 1 st -a-v -e-n -e-p 2 nd -a -ä-hpen -e-hpit 3 rd -a -ä-ba -e to imperative or tense-marked forms, is treated as such in the present study. However, three morphosyntactic aspects of potential mood verb forms make its classification as an inflectional category potentially questionable. First, verbs in the potential mood feature a segmentally separable marker (-tj-), rather than being part of a portmanteau morpheme simultaneously indicating mood/tense, number and normally person as is the case for other tenses and moods. Second, the stem allomorph chosen in all potential forms is consistently the weak form, which is quite consistent with the morphosyntactic behavior of other derived verbs which consistently have a specific consonant gradation type, while the mood and tense paradigms for non-derived verbs contain both strong and weak stem allomorphs. Finally, it is striking that the potential mood class marking suffixes and person/number suffixes (listed in Table 8.10) are homophonous with the class marking and present tense person/number suffixes for Class V verbs (cf. Table 8.23 on page 177). 10 In all of these three aspects, the potential forms of verbs are identical in behavior to a number of derivational verb forms (cf. §10.2.1, §10.2.2 and §10.2.3), and unlike other inflectional tense/mood forms. At this point, the only morphological motivation to classify the potential mood as an inflectional category is its complementary distribution with other tense and mood forms. These characteristics are summarized in Table 8.11 on the facing page. Table 8.11: Features of potential verb forms characterized as typical for inflectional or derivational forms consistent with features of potential forms inflection derivation consistently linearly segmentable marker ✓ consistently occurs with specific ∑-allomorph ✓ person/number marking like Class-V verbs ✓ complementary distribution with tense/mood forms ✓ With these facts in mind, potential forms could be analyzed as derived verb forms consisting of a lexical verbal root plus a verbalizer (the potential mood morpheme) followed by Class V inflectional suffixes. This possible analysis is illustrated in (31), in which the morphological components of the form gulatjav 'I will likely hear' are parsed and labeled.

Morphological marking strategies on verbs
In such an analysis, potential verbs no longer stand in opposition to tense and imperative mood forms, but instead are subject to a semantic restriction to a non-past time, and are thus only marked for present (i.e., non-past) tense, and are marked according to the present tense slots of the inflectional paradigm for Class V verbs.
It should be pointed out that the corpus contains insufficient data concerning the potential forms of any verbs in Class V. This is relevant because Class V verbs have bisyllabic stems that, together with the potential marker, may trigger allomorphy in other person/number suffixes, in which case not all potential forms would follow the standard Class V paradigm. Such information would be essential in fully evaluating the analysis proposed here. Due mainly to this lack of truly conclusive data, I continue to follow the standard classification of the potential mood as an inflectional category for the means of the present study, but point out this potentially problematic analysis for Pite Saami as described above as a topic worthy of future study.

Inflectional classes for verbs
Verbs in Pite Saami can be grouped into inflectional classes based on recurring patterns across inflectional paradigms. 11 Each verb is marked by a class suffix which is attached directly after the verb stem and precedes inflectional suffixes (cf. Figure 1 on page 149). Unlike nouns, the potential to have umlaut alternations and/or consonant gradation present for a given verb is dependent on the verb's membership in a specific class. However not every verb in the umlaut/gradation classes is subject to these alternations, as that is determined by whether the phonemes occupying the V1 position and the consonant center of the final foot, respectively, are susceptible to umlaut and/or consonant gradation. Furthermore, some derivational suffixes (such as the diminutive suffix -tj) can block consonant gradation and umlaut from occurring in the derived form. Membership in a specific verb class does not seem to be semantically motivated.
As described in the previous section, Pite Saami verb paradigms present complex combinations of linear morphology (inflectional suffixes) and non-linear morphology (consonant gradation, umlaut, vowel harmony), and consist of a minimum of 21 finite forms and several non-finite forms. This minimum includes 1 st , 2 nd and 3 rd person forms for singular, dual and plural in both present and past, as well as singular, dual and plural forms for imperative. 12 These are by far the most common forms in non-elicited data from the corpus. Furthermore, the three non-finite forms infinitive, connegative and perfect were also considered in determining inflectional classes. The non-elicited portions of the Pite Saami Documentation Project corpus are simply too limited to even come close to providing complete paradigms for even a single verb, and so a majority of the verb forms composing the paradigms for the current study are from elicitation sessions. Approximately 30 more or less complete verb paradigms were recorded, which provides sufficient data to posit five inflectional classes. However, the true extent and finer details of the morphophonological patterns found across verb paradigms in Pite Saami must be left to future study; it is possible that, with more research, more verb classes may result, or that the present classes may need revision. As a result, what follows must be considered of a preliminary nature.
There are five main criteria for positing five different verb classesː

Inflectional classes for verbs
• the regularity of the pattern of vowels occurring between the stem and inflectional suffixes (i.e., the class marking suffix) • the number of syllables in the infinitive form • the presence of deviant person/number suffixes relative to the other verb classes • V1 and consonant center stem allomorphy patterning throughout the inflectional paradigm (i.e., umlaut and consonant gradation) • whether some verb forms are subject to vowel harmony, and which slots trigger such vowel harmony To summarize these differences, it is sufficient to look at the class suffix and the syllable count in the infinitive form, the regularity of person/number suffixes across classes, the presence of consonant gradation ('C-grad') and umlaut, and the presence/absence of vowel harmony, 13 as illustrated in Table 8.12.
Class I is the least complex class, and is therefore dealt with first in §8.5.1, while classes II, III, IV and V are described in §8.5.2 through §8.5.5. §8.5.6 briefly discusses the possibility of the existence of other verb classes. The verb årrot 'be' and the negation verb are dealt with in §8.5.7 and §8.5.8. The final section (8.5.9) provides a brief summary of the verb classes, including a table listing examples from each of the verb classes.

Class I
Verbs in Class I are relatively simple, and characterized as follows: • a bisyllabic infinitive form • the class marking suffix is consistently -o • potentially subject to consonant gradation and umlaut, but not vowel harmony The verb viessot 'live, feel' is provided in Table 8.13 as an example. Other examples of Class I verbs include: årrot 'reside', gårrot 'sew', gähttjot 'tell', lávvlot 'sing' and såggot 'drown'.  Table 8.14 on the facing page summarizes the gradation pattern and class suffixes for Class I verbs. Note that umlaut alternations align with consonant gradation alternations.

vies-o-v viess-o-n viess-o-p 2 nd vies-o viess-o-bähten viess-o-bähtet 3 rd viess-o viess-o-ba viess-o pst 1 st viess-o-v vies-o-jmen vies-o-jme 2 nd viess-o vies-o-jden vies-o-jde 3 rd vies-o-j vies-o-jga viess-o-n imp 2 nd vies-o viess-o-n viess-o-t inf viess-o-t conneg vies-o prf viess-o-m
There are a number of verbs which seem to be marked by -u as a class marker in infinitive, such as gävdnut 'exist' and pruvkut 'use; usually do'. While the data in the corpus are incomplete, such verbs likely pattern in essentially the same way as the verbs mentioned above marked by -o, only they are consistently marked with -u as the class marking suffix.

Class II
The characteristics of verbs in Class II are: • a bisyllabic infinitive form with a class suffix -a or -å • potentially subject to consonant gradation, umlaut and vowel harmony For most inflected forms, the class marking suffix is consistent with the class marking suffix in the infinitive form; however, eight forms are assigned a specific class-marking vowel, as listed in Table 8.15. Class II verbs can further be divided into two sub-classes, based on the class marking suffix in the infinitive form: Class IIa is marked by a, while Class IIb is marked by å. The verb bassat 'wash' is provided in Table 8.16 as an example for a Class IIa verb, and bårråt 'eat' for a Class IIb verb in Table 8.17.

Inflectional classes for verbs
post-stem vowel is also å, just as the nouns in noun Class Id (also marked by å). Table 8.18 summarizes the gradation pattern, class suffixes and locations for vowel harmony for Class II verbs; here, V stands for the vowel which marks the infinitive form (a for Class IIa and å for Class IIb). Note that umlaut alternations align with consonant gradation alternations.

Class III
Verbs in Class III are characterized as follows: • a bisyllabic infinitive form with a class suffix -e • potentially subject to consonant gradation, umlaut and vowel harmony Twelve forms are subject to vowel harmony (the same six as for Class II verbs, plus six more). The verb basset 'fry' is provided in Table 8.19 on the next page as an example for a Class III verb. Other examples of Class III verbs include: vádtset 'go', adnet 'have, possess', diehtet 'know', båhtet 'come', buälldet 'ignite, burn' and máhttet 'can'. When the consonant center of a Class III stem consists of a single segment in the 1sg.prs and 2sg.prs forms and the V1 vowel is neither á nor ua/uä/uo, the class marking vowel is á instead of a, as illustrated in Table 8.19. Table 8.20 on the following page summarizes the gradation pattern, class suffixes and locations for vowel harmony for Class III verbs. Note that umlaut alternations align with consonant gradation alternations.

Class IV
Class IV verbs are characterized by: • a bisyllabic infinitive form • no allomorphic variation in the stem and in the class marker • deviant person/number suffixes with a -j-element The stem and the class marking suffix are consistent in all forms throughout a paradigm, i.e., there is no allomorphy in the stem or class marker. The person/number suffixes for 3sg.prs, 1du.prs, 3pl.prs, 1sg.pst, 2sg.pst and 3pl.pst deviate from the corresponding person/number suffixes in other verb classes in featuring an initial -j-element. A nearly complete paradigm for the verb välldut 'marry' is provided in Table 8.21.
välld-u-bähten välld-u-bähtet 3 rd välld-u-ja välld-u-ba välld-u-je pst 1 st välld-u-jiv välld-u-jmen välld-u-jme 2 nd välld-u-je välld-u-jden välld-u-jde 3 rd välld-u-j välld-u-jga välld-u-jin imp 2 nd n/a n/a n/a inf välld-u-t conneg välld-u prf välld-u-m However, the data in the corpus are not nearly sufficient to provide much more than the paradigm in Table 8.21. Class IV is likely a relatively small class of verbs; other potential candidates are årrat 14 'fall asleep', ádnot 'request' and tjerrot 'cry'. Lehtiranta (1992: 154) includes a paradigm for tjerrot, which appears to pattern like välldut. 15 The class marking vowel in the infinitive form is thus not restricted to the -u-indicated in Table 8.21. Table 8.22 summarizes the preliminary class suffix pattern for Class IV verbs, as well as the presence of a person/number suffix which deviates from the corresponding person/number suffixes in other verb classes. This is based on the paradigm for välldut in Table 8.21 on the previous page and the paradigm for tjerrot provided in Lehtiranta (1992: 154). 16

Verbs
n/a n/a n/a n/a n/a n/a inf

Class V
Verbs in Class V are characterized by: • a trisyllabic infinitive form with the class marking suffix -i • absence of consonant gradation, umlaut and vowel harmony Many Class V verbs are derived verbs based on a bisyllabic verb (cf. gullat 'hear' and gulladit 'be in touch' (lit.: let someone hear from you)). 17 The paradigm in Table 8.23 on the facing page provides an example for the verb ságastit 'speak'; other Class V verbs include bargatjit 'work a little ', gatjadit 'ask', gullalit 'listen', málestit 'cook, boil', gávnadit 'meet' and leradit 'teach'. Table 8.24 then summarizes the gradation pattern and class suffixes for Class V verbs.

Inflectional classes for verbs
-a-8 Verbs

Other possible verb classes
The data in the corpus are unfortunately not sufficient to be entirely confident concerning the five inflectional classes for verbs proposed here. With this in mind, the data concerning several verbs seem unusual, but also contradictory and inconsistent. Specifically, limited data on the verbs årret 'sleep', årrat 'fall asleep' and ádnot 'request' exist in the corpus indicating that these may belong to Class IV or some subset of Class IV verbs. Furthermore, a number of verbs with bisyllabic infinitive forms marked by -i-as a post-stem class-marking suffix exist in the data in the wordlist compiled by the Wordlist Project (cf. §1.2.3.4); however, in many cases, it seems that these verbs in fact belong to Class III, and the -i-class marker is simply an inconsistent spelling of e, as the realizations of /i/ and /e/ in unstressed syllables are more centralized, and thus easily confusable, particularly when applying what are otherwise Swedish graphemes representing more distinctly front Swedish vowels. For instance, the verb virrtit 'must' should perhaps be spelled virrtet and likely belongs to Class III. More data on this and other bisyllabic verbs with the -i-spelling need to be gathered to determine whether another inflectional class exists, or if these are only subclasses for Class IV and perhaps Class I, II or III.

The verb årrot 'be'
The verb årrot 'be' can be used both as a copula (cf. §13.1.4) and as an auxiliary (cf. §13.1.5.2); its paradigm is presented in Table 8.25. It is an unusual verb in a number of ways; these are listed on the following page. n/a n/a n/a inf årrot conneg lä prf urrum/lam 8.5 Inflectional classes for verbs • årrot 'be' is suppletive, featuring the two stems årr-and l-.
• Many of the l-stem forms are monosyllabic.
• The 2sg.prs, 3sg.prs and 3pl.prs forms can be shortened to 'l and encliticized onto the preceding word of an utterance if the preceding word has an open final syllable, as in (32) • The 1sg.pst form lidjiv is often shortened to lijiv, and the 3pl.pst form lidjin is often shortened to lin.
• The infinitive and perfect forms are the only forms in this basic paradigm which use the årr-stem, which is homophonous (and cognate) with the verb årrot 'reside, live'.
• Finally, the verb årrot 'be' is unique in having a contracted connegative and perfect form: lam 'be-prf\conneg' is a shortened form of lä 'be\conneg' and urrum 'be-prf', and is thus only used in conjunction with the verb of negation, as illustrated by the example in (33).

The negation verb
The negation verb is unique because it only exists as a finite verb; thus there are no non-finite forms. Table 8.26 on the following page presents the paradigm for the negation verb. Concerning the imperative forms, both forms indicated for each number slot are attested in the corpus.  Table 8.27 on the next page is provided to facilitate a cross-class comparison of inflectional paradigms with examples from the various inflectional classes for verbs, as well as the verb årrot 'be' and the negation verb. While the whole paradigm for each word is not listed due to a lack of space, the forms for inf, 2sg.prs, 3sg.prs, 2sg.pst, 3sg.pst and conneg are sufficient to convey the relevant morphological differences between the classes.  IV välld-u-t välld-u välld-u-ja välld-u-je välld-u-j välld-u 'marry'

Other word classes
This chapter describes the word classes: • adverbs in §9.1, • adpositions in §9.2, • conjunctions in §9.3, • and interjections in §9.4 The information provided here is of a preliminary nature due to limited data in the corpus, and stands to gain much from future research.

Adverbs
Adverbs compose an open word class and are defined by their ability to head an adverbial phrase; they can be further divided into two main groups: • derived adverbs

• lexical adverbs
Here, §9.1.1 deals with the former, while §9.1.2 presents with the latter.

Derived adverbs
At least one derivational affix seems to exist which derives an adverb from an adjective: the suffix -t, as illustrated by

Lexical adverbs
A group of lexical items exclusively used as adverbs in Pite Saami forms a subset of adverbs. A list of some lexical adverbs is provided in Table 9.2 on the next page. 2 Examples containing the sentence adverbs ber 'only', kan 'maybe', aj 'too' and vanj 'definitely' are provided in (3)  A further lexical adverb is gal 'actually', which can be used to emphasize a contradiction or surprise, as in (7). The interjection nä 'no', a borrowing from Swedish, 3 is also used in this example, in addition to the native Saamic negation verb.  allowed\conneg 'No,actually no, In (8), the adverb ilá 'too' modifies the adjective nuora 'young'.

The question marker gu∼gus
In several Saami languages, including closely related Lule Saami, a grammatical unit often referred to in the literature as a 'question particle' is used to mark polar interrogative clauses. 4 For Pite Saami, Lagercrantz (1926: 20-21) indicates that Pite Saami also has a question marker gu identifying polar interrogatives, although he shows that it is not obligatory. 5 In the entire Pite Saami Documentation Project corpus, there are only three clear tokens of a polar interrogative with the question marker, and even then, the marker has two forms: gu and gus. These tokens are provided in examples (9) 6 through (11). Based on this lack of data, and on the description provided in Lagercrantz (1926), one can only conclude that the question marker is no longer required to identify polar interrogative clauses, and has all but disappeared from current Pite Saami usage.
In determining which word class the question marker belongs to, several facts should be considered. Most importantly, like the adverbs in examples (3) through (6) above, the scope of the question marker is the entire sentence; gu∼gus indicates an epistemic lack on behalf of the speaker concerning the proposition expressed by the interrogative clause it marks. While its monosyllabicity is remarkable, and implies a strong degree of grammaticalization (since lexical items in general are minimally bisyllabic), a number of other monosyllabic lexical adverbs also exist (cf. Table 9.2 on page 185). On this basis, the question marker can be classified as a lexical adverb.
However, although the data are much too limited to be certain, the question marker in all three examples occurs directly after the finite verb. If it indeed can only occur here, then this may be sufficient reason to consider the question marker to be the sole member of a unique word class (perhaps best named 'particle') defined by its clause-level scope and syntactic position restriction. 7

Adpositions
Adpositions in Pite Saami constitute a closed class of words that are defined syntactically by their ability to head an adpositional phrase (abbreviated 'PP' as these are either postpositional or prepositional phrases). Postpositions, which are clearly preferred over prepositions, are covered in §9.2.1. The limited data on prepositions, which, with one exception, can all be used as postpositions as well, are described in §9.2.2. Table 9.3 provides a selection of postpositions found in the corpus and includes English translation equivalents. It is possible that other postpositions also exist but were not attested in the corpus.

Prepositions
With the exception of dugu 'like', which governs a noun in either the essive or the nominative case (cf. §5.2.9), a few words that are normally used as postpositions may also occur as prepositions. The corpus provides only a very limited amount of data concerning the existence and behavior of prepositions; the two examples are presented here.
In (14), birra 'about, around' is used as a preposition, and governs the genitive case on the complement demonstrative. birra birra around danne d-a-nne dem-dist-gen.sg 'and ribbon, purchased ribbon around that' [pit080708_Session08.012] In (15), badjel 'over' is used as a preposition. However, the complement nällje kronor 'four crowns' (referring to the Swedish currency) consists of the Pite Saami numeral nällje and a Swedish borrowing kronor which is inflected according to Swedish grammar (kron-or 'crown-pl'), and not Pite Saami grammar, so it is impossible to know with these data which case badjel governs as a preposition.  [pit0906_Ahkajavvre_a.159] Note that there are numerous examples for both birra and badjel as postpositions. It is not surprising that prepositions are infrequent and marginal in Pite Saami as other Saami languages also only have a small set of prepositions with significant restrictions in frequency and meaning. 8

Conjunctions
Conjunctions in Pite Saami form a closed class of words that connect phrases or clauses. A list of some conjunctions, what they can connect, as well as their English translation equivalents can be found in Table 9.4. Note that the conjunctions att, eller and men are borrowings from Swedish.
Conjunctions connecting clauses are discussed in §14.1 on coordination and §14.2 on subordination. Conjunctions connecting phrases are briefly described here. NPs, APs and verbs can be connected to another phrase of the same type by a conjunction; however, it is not clear from the data whether PPs or AdvPs can be connected. Some examples can be found in (16) through (21).

Interjections
An interjection is an individual word that is syntactically an utterance of its own at the same level as entire clauses. As such, interjections are not a part of another clause. Interjections often indicate a speaker's feelings or attitude towards an event. The data in the corpus are quite limited, and it is beyond the scope of the current study to describe interjections in detail, so the list of interjections and their English translation equivalents provided in Table 9.5 is preliminary and subject to amendment pending future study. Nonetheless, examples from the corpus of nå and jå are provided below. Note that the interjections jå, nä, så, å/oj, mmm and jaha are borrowings from Swedish. The interjection nå 'well, yes, ok' is very common in the corpus. It has at least three possible meanings. At the beginning of the conversation presented in (22), nå is a kind of declaration that a speaker is beginning to speak. As in the final utterance in this example, nå also indicates a switch to a new topic. The interjection nå can also be a confirmation of the preceding utterance, as in (23) Finally, nå can be used to indicate that a speaker is finished speaking, usually at the conclusion of a narrative and after a pause. One example can be found in the narrative in the recording 'pit100404' between utterance '.324' and utterance '.361'; due to space constraints, only the last two utterances of this long narrative are presented in (24)

Derivational morphology
Pite Saami is rich in derivational morphology. While it is beyond the scope of the present work to provide a thorough description of all the various derivational processes and of their semantic nuances and productivity, 1 the following should provide a general impression of how derivational morphology works in Pite Saami, as well as an overview of some of the more common derivational morphemes attested in the corpus and extant in the wordlist compiled by the Wordlist Project (cf. §1.2.3.4).
In the following, derivational meanings are assigned to suffixes for simplicity in classification; however, as with inflectional suffixes, derivational suffixes coincide with non-linear morphology when the derivational base is subject to non-linear morphological alternations. There are many nominalizing and verbalizing derivational processes, and derivations can apply to already derived forms. On the other hand, there are only two adjectivizers and one adverbializer.
Nominal derivation and verbal derivation are especially complex because the semantics of a derived word do not consistently equal the sum of the meanings of its components. Furthermore, the borderline between polysemy and homonymy of suffixes cannot always be clearly determined, and the decision whether two formally identical, but semantically different forms should be ascribed to the same morpheme or to distinct morphemes is not always obvious. This is reflected in the glossing standards used here in which most nominalizers and verbalizers are simply allotted numbers, as in nmlz1 or vblz3, as opposed to more meaningful glosses such as dim.
In the following, nominal derivation is dealt with first, in §10.1, before moving on to verbal derivation in §10.2, while adjectival and adverbial derivation are described briefly in §10.3 and §10.4, respectively. The final section ( §10.5) provides a summary of the derivational morphemes discussed here.
Note that examples in the present chapter include references to either the documentation corpus or an entry in the database from the Wordlist Project (cf. §1.2.3.4). Nearly all references to the documentation corpus are for elicita-tion sessions, and these are marked accordingly. Unlike examples in the other chapters, in which references indicate a particular utterance of a recording, references here may not be not more specific than the recording name alone because the relevant data were obtained during the course of a longer discussion, and not just in a single utterance. References referring to the Wordlist Project's database consist only of the four-digit entry number.

Nominal derivation
Nouns can be derived from verbs, adjectives, or other nouns. Some of the more common derivational suffixes are -tj, -k, -o, -däddje, -vuohta, and these are discussed in the following sections. The bases they can be applied to are summarized in Table 10.1.

The diminutive suffix -tj
The diminutive suffix -tj (glossed as dim) can be affixed to a nominal base to form a denominal noun with a diminutive meaning. Examples can be found in (1) through (6).
( The diminutive form features the same stem found in the nom.pl form of a noun paradigm (differences in the segments occurring between the consonant center and the right edge of the nominal base in these examples are due to regular alternations in inflectional noun class suffixes on the base). The resulting diminutive nouns are class IIIa nouns; a nearly complete paradigm for the derived noun guolátj 'little fish' is provided in Table 5.20 in §5.4.3. Note also that there is a diminutive verbalizer suffix -tj; cf. §10.2.1.

The general nominalizer suffix -k
The nominalizer suffix -k (spelled -g-intervocalically; glossed as nmlz 2 ) can be affixed to a noun, a verb or an adjective. The resulting derived nouns have a wide variety of meanings, but generally have a referent which is someone or something with a property referred to by the base. A number of examples are provided in (7) through (17) below, but this is hardly an exhaustive sample. In (7) the derived noun guhkajuolgagijd 'long-legger-acc.pl' is based on a compound noun guhka-juällge 'long-leg', and is used several times in the corpus to refer to moose.

The action nominalizer suffix -o
The nominalizer suffix -o (glossed as nmlz1) can be affixed to a verbal base to form a deverbal noun. In general, the resulting noun refers to the action or the result of the action denoted by the stem, as in (18)  However, as (24) and (25) indicate, the deverbalized noun does not have to refer exactly to the action or result of the verb, but only to a related concept.

The agent nominalizer suffix -däddje
The nominalizing suffix -däddje (glossed as nmlz2) creates an agent noun, indicating that the referent of the noun is involved in the activity denoted by the base. Examples are provided in (27)

Nominal derivation
The base is typically a verb, but can be a noun, as in (30). The stem of the derived agent noun is in the weak grade. As illustrated by (28) and (29), the resulting agent noun (with a root máles and åhpa) may no longer be directly derivable from the comparable verb (there is no verb *málle-t 'cook-inf', only málestit, nor a verb *åhpa-t, but instead åhpådit 'teach-inf').
Note that the noun báhkojådedäddje 'chairperson' in (29) is a compound calque based on the Swedish equivalent ordörande, which literally means 'word-driver'. It is not clear whether ? jådedäddje 'driver' exists on its own.

The state nominalizer suffix -vuohta
The nominalizing suffix -vuohta (glossed as nmlz3) typically derives nouns from adjectives, as in (32)  The suffix -vuohta can also be applied to a derived adjectival base. In (35), the stem máhtelis 'possible' is itself a derived adjectival based on the verb máhttet 'can'. The deepest derivational base in the example in (36) is roughly analogous, but one step farther removed from the final derived form: the highest-level base bargodis 'unemployed' is an adjectival form of the noun bargo 'work', which itself is a deverbal form based on the verb barrgat 'work' (cf. example (18) in §10.1.3 above).

Verbal derivation
Verbal derivation in Pite Saami is a particularly complex area, and the interested reader is first and foremost referred to Israel Ruong's PhD thesis Lappische Verbalableitung dargestellt auf Grundlage des Pitelappischen 4 (Ruong 1943). This work presents a comprehensive typology of non-derived verbs and verbal derivation suffixes in Pite Saami. It includes an extensive semantic sub-classification of the derivational suffixes into the varied and overlapping meanings each one can have. The forty suffixes Ruong presents, and the myriad functions he assigns them to, further attest to the complicated nature of verbal derivation in Pite Saami. The present discussion cannot hope to improve on Ruong's work, and instead attempts to use the Pite Saami Documentation Project corpus to achieve the following: • Using the diminutive verbalizer -tj as a starting point, illustrate the complexity of verbal derivation in Pite Saami due to the persistent irregularities between forms and functions ( §10.2.1); • Present a sample of verbal derivations ( §10.2.2 through §10.2.4); • Provide a basic description of the important valency-decreasing verbal derivation creating passive verb forms ( §10.2.5).

The diminutive verbalizer suffix -tj and the complexities of Pite Saami derivational verb morphology
The diminutive verbalizing suffix -tj (glossed as dim) expresses doing the activity referred to by the verbal base a little bit or to a limited extent, as in (39) through (41). 5

Verbal derivation
walk-dim-inf [2047] (41) bällke-t → bielka-tji-t 'have a small quarrel' quarrel-inf quarrel-dim-inf [4698] The weak form of the base verb is selected by -tj, and the final vowel in the base becomes a. The i following the -tj suffix is the verb class marker for the resulting Class V verb.
Note, however, that other derivational suffixes can produce diminutive meanings as well, as illustrated by the examples in (42) through (44).
cough-vblz2-inf [4898] (44) rassjo-t → rässjo-dalla-t 'rain lightly' rain-inf rain-vblz3-inf [5073] In these three examples, the derivational suffixes -st, -d and -dall, respectively, 6 also derive deverbal verbs which add similar diminutive meanings to the base. If these suffixes were restricted to a diminutive meaning, then this would simply be a case of many forms corresponding to a single function. However, these suffixes, which are all quite common, only occasionally carry a diminutive meaning. In other instances, they impart a variety of different meanings to the base form. This is illustrated by just a few examples below, and is even more obvious throughout Ruong (1943). Despite the variety of and inconsistencies in the meanings that verbal derivational suffixes express, their limited number relative to the number of functions they fulfill is reason enough to describe each of these suffixes as a single derivational affix with multiple functions, rather than multiple, homonymous affixes, each aligned to a separate function.

The verbal derivational suffix -st
In addition to the diminutive meaning in (42) above, the derivational suffix -st (glossed as vblz1) is applied to a postposition in (45), and functions as a verbalizer. In (46), the nominal base is not only verbalized, but has a causative or perhaps an inchoative meaning. The derived verb in (47) is a figurative extension of the verbal base's meaning. Furthermore, -st can indicate that an action is carried out briefly or for a short period of time, as in (48) (48) basse-t → base-sti-t 'fry quickly' fry-inf fry-vblz1-inf [5501] Note that the i following the -st suffix is the verb class marker for the resulting Class V verb.

The verbal derivational suffix -d
In addition to the diminutive meaning in (43) above, each of the two examples of the verbalizer -d (glossed as vblz2) in (49) and (50) has a reflexive meaning; note that the base in (50) is a noun, not a verb. The example in (51) has a transitivizing effect on the verbal base, while there is no clear difference in meaning between the base and the resulting derived form in (52) and (53). The last example, sykel 7 'bicycle' in (54), illustrates that this suffix is quite productive, as it is used as a verbalizer for a loanword serving as a nominal base. Note that the i following the -d suffix is the verb class marker for the resulting Class V verb.

The verbal derivational suffix -dall
In addition to the diminutive meaning in (44)  Passivization is a valency-decreasing device because the resulting verb is intransitive, as it only features the patient-like argument as its sole core argument in nominative case. Note that Svonni (2009: 92) claims, for North Saami, that "one cannot indicate the agent in any way" (my translation) in passive clauses using the cognate North Saami passivizing suffix. Pite Saami differs significantly from North Saami in this respect, as Ruong -himself a native speaker of Pite Saami -verifies (cf. Ruong 1943: 41). It is very possible that the Pite Saami strategy of placing the agent in an oblique case could be due to extensive language contact with Swedish, a language which clearly allows the agent in a passivized clause to be expressed obliquely using a prepositional phrase headed by the preposition av 'of, from'. Indeed, Swedish PPs headed by av in other contexts are best translated into Pite Saami as an NP in elative case, the same oblique case in which the agent NP in a passive Pite Saami sentence is found.
Some other examples of transitive verbs and their passivized equivalents using -duvv are shown in (62) through (64).
utilize-pass-inf [2682] There are not sufficient data in the corpus to state any more about passive derivation, particularly concerning morphophonological effects of passivization 10.3 Adjectival derivation on verb stems, and this and other related topics must be left for future study. The reader is referred to Ruong (1943) for a more thorough morphological and semantic account of Pite Saami passives. Inflectional aspects of passivized verbs are treated in §8.3, while syntactic aspects of clauses with passive verbs are presented briefly in §13.1.1.1.
Note that the derivational suffix -duvv can have meanings other than passive when attached to a nominal or adjectival base. Typically it then expresses a change of state that is related to the referent of the root involved. A few examples are provided in (65)

Adjectival derivation
Only two derivational processes exist for adjectivals: the non-productive derivation of adjectives by -s, and the productive derivation of ordinal numerals from cardinal ones. These are described below.

Adjective derivation
It seems conceivable that adjectives can be derived by the suffix -s (glossed as adjz). For instance, bahá is a nominal meaning 'evil', as in (69), and bahás is the equivalent attributive adjective form, as in (70) In addition, the nominalized form bahá-k 'evil' can be further derived into an adjective bahágis 'painful' as in (71).
However, as pointed out in detail in §7.1 through §7.3, not all adjectives follow this pattern. In fact, based on the current data, the -s suffix marks attributive adjectives (as in (70)) and as well as predicative adjectives, and, synchronically, it is not considered to be productive for either attributive or predicative forms at all.

Ordinal numeral derivation with -át
Numerals, a sub-category of adjectivals (cf. §7.9), are subject to derivation. The basic ordinal numerals can be derived by applying the derivational suffix -át (or its allomorph -et) to the respective cardinal numeral, although the forms vuostas 'first' and mubbe 'second' are suppletive. Ordinal derivation is discussed in §7.9.1.1 in more detail, including a comparison of cardinal and ordinal numbers in Table 7.7 on page 144.

Adverbial derivation
Adverbs are not common in the corpus (as opposed to other word classes and phrase types with adverbial functions), but do appear to be derivable from an adjective base using the suffix -git. This is dealt with in more detail in §9.1.1. Table 10.2 on the facing page provides an overview of the derivational morphology discussed in this chapter.

Phrase types
There are five types of phrases in Pite Saami which form syntactic constituents of other phrases or of clauses: • verb complex (VC) • nominal phrase (NP) • adjectival phrase (AP) • adverbial phrase (AdvP) • postpositional phrase (PP)

Verb complex
The Pite Saami verb complex (abbreviated 'VC') consists minimally of a finite verb, and maximally of a finite verb and one or two non-finite verb forms. With the exception of the imperative, the finite verb inflects for tense or mood, number and person, and agrees with the subject. The imperative only inflects for number.
In combination with non-finite verb forms, the verbal categories negation, mood, and aspect can also be expressed.
To better describe the distribution of finite and non-finite verbs forms in VCs, verbs are divided into two groupings: • lexical verbs and the copular verb årrot 'be' • grammatical verbs (the negation verb, the aspectual auxiliary verb årrot 'be', and the modal verbs; cf. §13.1.5).
In VCs featuring only one verb form, the finite verb is a lexical verb or the copular verb. In VCs with two or three verb forms, the finite verb is a grammatical verb, while the selection of each non-finite form is determined by the type of verb governing it: the verb of negation triggers the connegative form, the aspectual auxiliary verb triggers either the perfect or the progressive form, and the modal verbs trigger the infinitive form. This is summarized in Table 11.2. The constituent order of the individual verbal components is not strictly set, although the ordering indicated in Table 11.2 is most common. Furthermore, other clause-level components may occur between these verb forms (cf. §12.2 and §13.1.5).
For instance, the examples in (1) and (2) each feature a VC consisting solely of a finite verb. In (1) it is the singular imperative form of the lexical verb vädtjat 'fetch', while in (2) it is the encliticized 3sg.prs form of the copular verb. The corpus does not provide any evidence for three-verb VCs with a modal verb as the finite verb, so whether this logically possible structures is acceptable must be left to future research. However, it is clear that the negation verb is only attested as a finite verb, and can never occur as the second or third verb in a multi-verb VC.

Nominal phrases
Nominal phrases (abbreviated 'NP') in Pite Saami are divided into two groups: • full NPs

• pronouns
NPs can function as arguments, adjuncts, predicates, adverbials, dependents of postpositions and possessors or modifiers of other NPs. They consist of at least one nominal component that inflects for case and number. Note that NPs can also be modified by postpositional phrases and non-finite verb forms, but due to a lack of sufficient data, a description of these NP modifiers must be left for future study. Finally, relative clauses also modify an NP; these are covered in §14.2.4.
NPs have the structure illustrated in (10), with optional components in parentheses.
(10) [(demonstrative) + (other modifier(s)) + nominal + (refl-intensifier) + (rel-clause)] N P Either a noun or a pronoun forms the head of an NP. The demonstrative, the nominal and any attributive reflexive pronoun always inflect for case and number, while generally the other modifiers do not. Other modifiers may be an adjectival phrase, a numeral or an NP in genitive case. An intensifier in the form of a reflexive pronoun (cf. §6.3) can modify a noun phrase as well, and occurs after the head. Finally, the head can be modified by a relative clause, which also occurs after the head. Some examples for possible NP structures are found below.
The only NP in (11) consists solely of the noun Tjeggelvasav 'Lake Tjeggelvas'.

Tjeggelvasav
[Tjeggelvasa-v] N P Lake.Tjeggelvas-acc.sg 'And then I see Lake Tjeggelvas. ' [pit100404.013] In the example in (12), the NP dat ello 'that reindeer herd' consists of a demonstrative and the head noun, and is the subject of the clause. In (13), the subject NP consists of the head noun ello 'reindeer herd' and the genitive NP dáj Skailej 'of these Skailes' 2 which modifies the head noun. An NP marked for genitive case can also function as a modifier that narrows the reference of the head noun by signifying some characteristic of the head noun's referent, as in (13) [pit080924.253] In the example in (15), the NP nåv edna båtsoj 'so many reindeer' 3 consists of the AP nåv edna 'so many' and the head noun båtsoj 'reindeer'. As the example in (16) illustrates, it is possible for more than one modifier to be included in an NP. Here, both APs guäkte 'two' and stuor 'big' modify the noun head, which is the derived compound guhkajuolgagijd 'long-leggers' (referring to moose). If the context is sufficiently clear, it is possible that the head noun is not realized when a demonstrative and/or modifier is present; such cases are referred to as elliptical constructions. Numerals, for instance, can be in elliptical constructions, as in (17) With the exception of the acc.sg form of akta 'one', numerals do not inflect for case and number (cf. §7.9).

Nominal phrases
Less commonly, an adjectival phrase can be in an elliptical construction, either with or without a demonstrative (cf. §7.1.1). In the absence of a head noun, the adjective in the AP is the host for case and number, and is morphologically a nominal. In the example in (18) includes a demonstrative; here, the head noun referring to a girl is not overt, and the NP only contains the demonstrative dat 'that' and the adjective tjábba 'beautiful'. The sentence in (19) illustrates a similar structure, but without a demonstrative. In this case, the head noun referring to a piece of candy is not realized, and the NP consists only of the adjective rupsisav 'red', which inflects for case and number. 4 The second NP in (20), såbe 'stick', also illustrates a genitive NP functioning as the complement of a PP.
When the head of an NP is a personal pronoun, it is also possible to modify it. For instance, in (22), the 1sg.nom reflexive pronoun etj modifies the 1sg.nom personal pronoun mån as an intensifier. Note that this ordering (modifier following the head) deviates from the general pattern, in which the modifier occurs before the head of the NP.

NPs in adverbial function
Nominal phrases, particularly when referring to time or place, and thus inflected for one of the locative cases (illative, inessive or elative), are often used as temporal or locational adverbials, as in giesen 'in summer' in (24), and in hiejman 'at home' in (25), respectively. Furthermore, the word vahkov 'week' in (25)

Adjectival phrases
Adjectival phrases (abbreviated 'AP') in Pite Saami are divided into three groups based on the type of adjective required as head: • attributive APs (headed by attributive adjectives) • predicate APs (headed by predicative adjectives) • numeral APs (headed by numerals) In the first case, the attributive AP is syntactically embedded in an NP whose head noun it modifies. In the second case, the predicative AP ascribes the property it concerns to the entire NP which is the subject of the copular clause which the AP is embedded in. Numeral APs can occur both attributively and predicatively.
The three possible constituent structures of APs are presented in (27) In (30), the predicative adjective nuorra 'young' agrees with the subject NP mánná 'child' in number.

Postpositional phrases
A postpositional phrase (abbreviated 'PP') is headed by a postposition, which is always preceded by an NP complement. Any components in this complementing NP which are subject to case inflection inflect for genitive (as well as number). This structure is illustrated in (40). See Table 9.3 on page 188 for a list of postpositions.

(40) [NP [gen] + post-position] PP
The complement in a PP can be any valid nominal phrase. A number of examples for various NPs complementing the head of a PP are provided below: a noun with a demonstrative in (41), a single noun in (42), a demonstrative pronoun in (43), a personal pronoun in (44), an interrogative pronoun in (45)

Overview of the syntax of sentences
In describing Pite Saami clauses, it is useful to begin with basic clauses that contain a full predicate and its arguments, complements and/or adjuncts, before moving on to describe complex clauses which consist of two or more clauses linked to one another. Therefore, basic clauses are described in Chapter 13, including declarative, interrogative and imperative clauses. Chapter 14 then deals with complex clauses, covering coordination and subordination.
However, in order to better understand the syntax of sentences, it is sensible to begin with two general discussions that provide a framework for understanding the syntactic descriptions that follow. The first of these, in §12.1 below, covers grammatical relations in Pite Saami. This leads to the second discussion in §12.2, which concerns clause-level constituent ordering, and the likely role that information structure plays in determining this.

Grammatical relations
Pite Saami is an accusative language because the only argument of an intransitive verb (S) is marked in the same way as the most-agent-like argument of a transitive verb (A): by the nominative case. The most-patient-like argument of a transitive verb (P) is marked differently: by the accusative case. This is illustrated by the following examples, with an intransitive verb in (1) and a (mono-)transitive verb in (2) [pit100404.216] The direct object of a ditransitive verb is also in the accusative case, while the indirect object of a ditransitive verb is in an oblique case (usually in the illative case, which prototypically indicates that the noun refers to the goal of a movement), as illustrated in (3) The following section provides more examples illustrating the flexibility of constituent ordering.

Constituent order at clause level
Clause-level constituent ordering in Pite Saami is not determined syntactically. That being said, in elicited clauses from the corpus, some ordering patterns do occur more frequently than others, and indicate that SVO ordering is preferred in context-free elicited clauses, everything else being equal. This is illustrated by the examples in (5)  It is possible that this ordering is triggered by typical Swedish constituent ordering, which is generally SVO, as Swedish was often used as the meta-language in elicitation sessions.
More significantly, the part of the Pite Saami corpus consisting of natural language situations confirms the lack of any set constituent ordering based on syntactic criteria. 1 To illustrate this syntactic flexibility, examples of SOV and OSV constituent ordering are provided in (8) and (9) [pit090702.149] The example in (10) has VSO constituent order, and additionally has the nonfinite verb complement (with OV ordering) in clause-final position.
1 Note that Sammallahti claims that at least for North Saami (although it is not entirely clear whether he means North Saami or is generalizing for all Saami languages here), the "order of the main constituents […] is largely free from formal restrictions and guided by pragmatic principles", but then states that the "basic order is SVO" (Sammallahti 1998: 95). This seems to reflect the data from the Pite Saami corpus to the extent that context-free elicited clauses tend to be SVO, while in fact no syntactic criteria for constituent ordering can be ascertained in natural language. Lagercrantz takes several pages to describe a variety of tendencies in constituent ordering for Pite Saami declarative clauses, even after describing ordering preferences concerning topic and focus within a discourse and summarizing the actual situation by stating quite vaguely that the position of the subject in a clause has a 'certain stylistic effect' ("Die Stellung des Satzgegenstandes hat eine gewisse stilistische Wirkung") (Lagercrantz 1926: 46). Perhaps current descriptions of the syntax of the Saami languages would be better served if linguists would cease trying to force these languages into an inaccurate (but typologically neat) label such as SVO.
(10) dä dä then galgav galga-v will-1sg.prs mån mån 1sg.nom gåvåjd gåvå-jd picture-acc.pl vuosedit vuosedi-t show-inf 'Then I will show some pictures. ' [pit080825.036] Attempting to determine constituent order patterns is further complicated by the fact that it is sometimes impossible to tell what the constituent ordering is because NPs referring to information provided by context alone are frequently not realized overtly, as in (10) above and (11)  While person and number markers on the finite verb indicate grammatical information about the subject, there is no overt subject in (11) or (12). The clauses in (12) through (14) are lacking overt objects. The final example is also missing the indirect object.
Indeed, a complete clause can consist of nothing more than an inflected verb, as in the response in (15), which consists of nothing more than the copular verb inflected for 3sg.prs.
(15) A: ja ja and tjábba tjábba beautiful dállke! dállke weather\nom.sg 'and such beautiful weather!' comment involves a transitive verb, the object or complement clause (the focus) normally follows the verb, as in (6) and (7) above. However, clausal elements in focus can be moved from their 'default' position, which results in significant deviations from the preferred SVO constituent order. This is reflected in constituent interrogative clauses; here, the interrogative pronoun is in focus and always in clause-initial position (cf. §13.2.1).
The short example text presented in (19) and (20) below should serve to give an impression of how information structure may be the driving force behind constituent ordering at clause-level. Here, the speaker is talking about looking inside her mother's shoes after discovering that a mouse had been in them.  [pit100404.354] In the first clause (19), the topic is danne 'there', which refers to the shoes (the topic of the anecdote at this point) and is clause-initial. The constituent jåŋåtjav 'lingonberry' is the focus, but it is not particularly significant in the anecdote, and it follows the finite verb vuojdniv 'I saw'. However, when particular emphasis is placed on the focus, as in the following clause in (20), the constituent in focus can be fronted. Here, skafferijav 'pantry' is in focus, and receives particular emphasis 3 by occurring before the verb complex lä danne adnam 'has had there', while the topic (the mouse) is not realized overtly at all, but implied by the context and by the finite verb form inflected for 3sg. This fronting of a constituent is often accompanied by higher acoustic intensity, as is the case here.

Basic clauses
A basic clause is a syntactic unit at text-level consisting minimally of a finite verb. In declarative clauses and interrogative clauses, this finite verb is marked morphologically for person, number, tense and/or mood. Aspect can be expressed analytically at the clause level using an auxiliary verb and a non-finite verb form. In all basic clauses, the finite verb agrees in number and, with the exception of imperative mood, in person with the syntactic subject of the sentence, which is a nominal phrase in the nominative case. NPs referring to information provided by context alone are not necessarily realized overtly. As a result, the syntactic subject and other verbal arguments are often not overtly present.
The following sections first present basic declarative clauses with intransitive and transitive verbs, existential clauses, copular clauses and complex verbal constructions consisting of more than one verb ( §13.1). Then, §13.2 deals with interrogative clauses, before §13.3 and §13.4 cover syntactic aspects of the imperative mood and the potential mood, respectively.

Declarative clauses
Declarative clauses are the most common type of clause in the Pite Saami corpus. In the following, declarative clauses with a single verb are dealt with first, covering intransitive and transitive verbs, and two special cases (existential clauses and copular clauses). Then, declarative clauses featuring a modal or auxiliary verb in addition to the lexical head verb are described; because negation is expressed by an auxiliary verb, it is covered in the same section. While constituent ordering is mentioned in the following sections, it mostly refers to tendencies only, and the flexible nature of Pite Saami constituent ordering should always be kept in mind, as discussed in §12.2.

Basic intransitive declaratives
The subject of an intransitive declarative clause is in the nominative case, as in (1)  The complement of a copular clause can be an NP in the elative case in which case it describes the material which the subject referent is made of, as in (20) A copular clause can also function as an existential clause when it includes a temporal adjunct. In such cases, the existence of the subject referent is posited at that particular time indicated by the adjunct. Pragmatically, this usually announces an event connected to the subject referent. Typically, the temporal referent occurs first in the sentence, then the copular verb, and the subject is last (just as with the existential verb gädvnut; cf. §13.1.3), as it is usually the focus. This is illustrated by the example in (21).
(21) ja ja and dále'l dále=l now=be\3sg.prs káffa káffa coffee\nom.sg 'And now it's coffee (time). ' [pit090519.313] Possession can also be expressed by a copular construction. In such a construction, the possessed NP is the subject of the clause in the nominative case, which the finite verb agrees with in person and number. The possessor NP is in the inessive case. Such a construction is illustrated by the example in (22) In the corpus, such possessive constructions always have the constituent order possessor+copula+possessed. While this type of possessive construction is the native Saamic structure (Bergsland 1977: 9), it is very uncommon in the Pite Saami corpus, and almost exclusively limited to elicitation sessions. The elicitation scenario may have had an effect on the constituent order, 6 but it is more 13.1 Declarative clauses likely the case that the constituent order reflects information structure preferences, specifically the tendency for the topic (more often the possessor, which is animate) to come before the focus 7 (more often the possessed, which is inanimate).
In any case, a clause-level construction using the monotransitive verb adnet 'have' 8 expressing possession, as in (23)

Multi-verb declarative clauses
Verbs which govern non-finite verbal complements can be classified into three groups based on the type of non-finite complement verb form they co-occur with, as illustrated in Table 13.1. The finite verb occurs before the non-finite lexical

Interrogative clauses
Alternatively, the interrogative can be an adverb, as in (47) [pit080924.168] Assuming that any constituent which is the pragmatic focus can be marked by fronting, as preliminarily asserted in §12.2.1, then the fronting of the interrogative word is consistent with focus-marking. However, for constituent interrogative clauses, fronting is then obligatory. The rest of the clause is constructed syntactically just as freely as any declarative clause would be. While subjectverb inversion can occur, the flexible nature of Pite Saami constituent ordering prevents this from necessarily marking a clause as interrogative.
It is worth noting that the discourse marker nå, which can be translated as 'well' or sometimes 'yes', frequently precedes constituent interrogative clauses, as in (50). However, it is not obligatory, nor is it restricted to interrogative clauses. It is likely a discourse marker, perhaps simply indicating the speaker's active interest in the conversation.

Clauses in the potential mood
As mentioned in §8.1.3.2, the potential mood can also be used as a less severe command. This resembles clauses in the imperative mood by also never occurring with an overt subject, as shown in example (64).

Complex clauses
Two or more clauses can be conjoined by coordination or subordination. After coordination is covered in §14.1, complement clauses are presented in §14.2.1 and adverbial clauses are dealt with in §14.2.2. Finally, relative clauses which do not form a constituent of a matrix clause, but are instead part of a nominal phrase, are described in detail in this chapter as well (in §14.2.4).

Clausal coordination
There are several coordinating conjunctions that are used to syntactically join the basic clauses described in Chapter 13. In such cases, a coordinating conjunction occurs between the two clauses it connects. The clauses themselves are otherwise not marked in any way for coordination. The coordinating conjunctions are ja 'and', vala 'but', men 1 'but', jala 'or' and eller 2 'or'. The examples in (1) and (2) illustrate clausal coordination using the coordinators ja and men, respectively. (

Clausal subordination
both finite and infinitive predicates are possible. Complement clauses can be marked by a complementizer or can stand in juxtaposition to the matrix clause. The different complement clause marking strategies are summarized in Table 14.1 and described in the following sections.

Complement clauses with a finite predicate
Complement clauses with a fully inflected finite predicate are attested using one of two strategies. First, the borrowed complementizer att 3 can mark a complement clause. In such cases, the complement clause typically follows the matrix clause. The complementizer is in clause-initial position in the complement clause. Examples can be found in (4) and (5) Secondly, complement clauses with a finite predicate may be juxtaposed to the matrix clause they belong to. The complement clause typically follows the matrix clause. Verbs hosting such complements include jáhkket 'believe', diehtet 'know', hållåt 'say' and tuhtjet 'like'. Examples can be found in (6)

A grammar of Pite Saami
Pite Saami is a highly endangered Western Saami language in the Uralic language family currently spoken by a few individuals in Swedish Lapland. This grammar is the first extensive book-length treatment of a Saami language written in English. While focussing on the morphophonology of the main word classes nouns, adjectives and verbs, it also deals with other linguistic structures such as prosody, phonology, phrase types and clauses. Furthermore, it provides an introduction to the language and its speakers, and an outline of a preliminary Pite Saami orthography. An extensive annotated spoken-language corpus collected over the course of five years forms the empirical foundation for this description, and each example includes a specific reference to the corpus in order to facilitate verification of claims made on the data. Descriptions are presented for a general linguistics audience and without attempting to support a specific theoretical approach. This book should be equally useful for scholars of Uralic linguistics, typologists, and even learners of Pite Saami.