Diminutivizing L-reduplication in Norwegian

Abstract

This paper gives a broad overview of how Norwegian productively makes use of L-reduplication to convey diminutive meaning. I argue that this previously undescribed phenomenon is a diminutivizing process that copies the stressed vowel and any consonants until the next morpheme boundary. The construction can be attested as far back as the start of the 20th century and its realization varies geographically between two main variants. I show that L-reduplication is restricted phonologically, but applies productively (unlike other echo reduplicative processes) across different parts of speech.

1. Introduction

Norwegian speakers on occasion use forms like guttelutt (1b) and vovvelovven (1b). These forms are modified from the lexical roots gutt ‘boy’ and vovv ‘woof’. The systematic use, meaning and distribution of the modified forms are the object of study in this article. In a nutshell, I argue that such forms are best analyzed as diminutives (dim).

1. (1)

Norwegian is untypical among Indo-European languages because it is traditionally described as not having any productive morphological diminutive suffix. As a counter to this claim, this paper documents that one such process, making use of reduplication, has been present in Norwegian for at least 130 years (18a). Within the Indo-European language family, this strategy of forming diminutives through reduplication is typologically rare and more restricted than in Norwegian.

The phonological process by which gutt is modified to guttelutt and vovv to vovvelovven reduplicates the rhyme of the syllable that bears primary stress. The rhyming segment is re-added following a medial -el [əl], e.g. for guttelutt as given in (2). The diminutivizing morpheme can be formulated as -elV́C $_{1}$ , composed of the invariant segment el and a reduplicative segment whose realization is conditioned by the rule reduplicate V́C $_{1}+$ (read as: reduplicate the stressed vowel and one or more consonants until the next morpheme boundary). Because the consonant cluster that precedes the reduplicate vowel is seemingly ‘exchanged’ for [l], I refer to the phenomenon as L-reduplication.

1. (2)

Norwegian speakers often report that they intuitively associate L-reduplicate forms like guttelutt with child or animal-directed speech – ‘isn’t that how you would talk to a baby?’. This intuition might have some merit, but as this paper documents, L-reduplication is attested outside of these domains. The most straightforward evidence for this claim lies in the examples given in this article (appendix A is a list of every attestion of L-reduplication), which is found in exchanges on online forums that should not be understood as child-directed speech (3), or between adults in real-life unelicited contexts (4).

1. (3)

2. (4)

No previous work describes L-reduplication. On the contrary, a productive morphological diminutive is claimed not to exist for Norwegian. Faarlund, Lie & Vannebo (1997, p. 113) note explicitly that there are are no diminutive suffixes in Norwegian. The same claim is made in the first sentence of Skommer (2016, p. 1), which reads ‘mangler språket suffikser med diminutiv betydning.’ ‘the language [Norwegian] lacks suffixes with diminutive meaning’. Fløgstad & Eiesland (2019, p. 66) express the sentiment that there are no evaluative morphological markers that express diminutive meaning in Norwegian.

The present article presents a counterclaim: a productive morphological diminutive exists in Norwegian. The reader is encouraged not to take the claim of productivity as a claim of frequency; while L-reduplication is a phonologically regular and morphosyntactically productive phenomenon, its frequency is hard to ascertain and is intuitively lower than that of diminutives in languages where diminutives are frequently used, e.g. Dutch (see e.g. Gillis (1997, p. 8) for an impression of frequency) or Spanish (Montrul et al. 2013, p. 96). The aim of the article is to give a broad phonological and morphosyntactic overview of L-reduplication and provide examples of its usage.

The article is structured as follows. Chapter 2 discusses how Norwegian L-reduplication fits into the notion of diminutives (2.1) and then outlines crosslinguistic tendencies in reduplication used in diminution (2.2), the previous literature on Norwegian diminutives (2.3) and the history of L-reduplication (2.4) in Norwegian. Chapter 3 outlines the distribution of L-reduplication. I first discuss the frequency of L-reduplication (3.1). In section 3.2, I outline its diminutivizing properties in comparison with other rhyming strategies (3.2.1), how it is constrained phonologically (3.2.2 and 3.2.3) and how to best formalize the morphophonological rule that derives L-reduplication (3.2.4). Subsection (3.2.5) shows how certain lexical classes deviate somewhat from the most common L-reduplicative rule. Subsection 3.2.6 outlines variation in stress assignment in L-reduplicate forms across Norwegian dialects. In section 3.3, to underline the productivity of L-reduplication, I illustrate how L-reduplicated is found across various parts of speech and inflectional categories. Section 3.4 briefly discusses the distribution of L-reduplication in Danish and Swedish. Chapter 4 evaluates the place of L-reduplication in the morphological derivation (4.1) and the internal morphological structure of L-reduplicates (4.2). Chapter 5 summarizes the article.

2. Background

This section provides a context for the notion of diminutives and how Norwegian L-reduplication should be taken to be a diminutive (2.1). I then (2.2) discuss reduplication as a diminutivizing process crosslinguistically in a local, i.e. North Germanic, and global context. Lastly I give an outline of the literature on Norwegian diminutivization strategies (2.3) and of the history of Norwegian L-reduplication (2.4).

2.1. Diminutive semantics

A range of semantic functions are included under the umbrella term diminutive. Jurafsky (1996, p. 534) describes the variety of notions associated with diminutive exponents. The variety is well illustrated by the diachrony of the Spanish suffix -ita, which attached to ahor ‘now’ adds an intensifying component in Mexican Spanish and an attenuating force in Dominican Spanish. Jurafsky, by use of metaphor, justifies classifying both senses, although functionally different, under the umbrella term diminutive.

Certain functions are very commonly described across languages as diminutives. Among these are semantic concepts, e.g. smallness (5a), approximation (5b) and partitative semantics (5c), as well as pragmatic functions such as affection (5a), intimacy (5d) or derogation (5e). More thorough discussions of these categories and their subcategories are given by Jurafsky (1996), Rubino (2005) and Dressler & Barbaresi (1994).

1. (5)

The meaning of the Norwegian L-reduplicated diminutive is varied, but overlaps with the concepts outlined above: there are denotational associations with smallness or attenuation and force-modifying senses that may be broadly described as affective, cutesy, intimate, mocking or joking.

2.1.1. Smallness or youth

A Google search hints towards the smallness component. The lexeme guttelutt ‘boy.dim’ ¹ yields 56 results in Norwegian, 28 of which are part of a Norwegian utterance (rather than online usernames, single-word utterances, non-Norwegian contexts or contexts that pertain to an academic talk based on the present topic). Of these 28 sentences which contain variants of the lexeme guttelutt, 10 ( $\sim$ 36%) are preceded by (inflectional variants of) the adjective liten ‘little’. None are preceded by adjectives that denoate large size as ‘big’, ‘huge’ or ‘gigantic’.

Compare then a search for the un-reduplicated form gutt ‘boy’. The web-based corpus of Norwegian NoWaC (Guevara 2010) contains 114720 instances of the un-reduplicated lexeme gutt. 3010 ( $\sim$ 2.6%) of these are preceded by the adjective liten or its inflected variants. Things which are thought of as smaller might plausibly be understood to more often be described as small, compared with things which are not thought of as small. The difference in frequency of how often little precedes guttelutt compared to how seldom it precedes gutt may indicate that the former is usually thought of as smaller. Alternatively, because the Norwegian adjective liten is also used in reference to age, the finding indicates that a guttelutt is often thought of as younger.

2.1.2. Affection

The emotional component of endearment is similarly visible; 13 of the Norwegian 28 hits ( $\sim$ 46%) for guttelutt variants are preceded by affectional adjectives like sweet, adorable, perfect or cute. The only adjective with a negative connotation which is found together with guttelutt is stakkars ‘poor/pitiable’, which is used for pity rather than as a display of negative sentiment from the speaker.

2.1.3. Attenuation

The attenuating force of L-reduplicated forms may be read from (6). The author reduplicates the verbal infinitive jobbe to form jobbelobbe. ² A possible reading of (6) is that the writer intends not to work a full day or perform less intensive work-related tasks. A different but similar interpretation of diminutives is given for Greek by Sifianou (1992, p. 159) as markers of humility. Sifianou’s interpretation yields a similar and plausible reading of (6) where the writer tries to hedge their act of working in order to avoid the possibly self-aggrandizing label of someone who a. enjoys work or b. is willing to work even during the weekend.

1. (6)

2.1.4. Derogation

L-reduplication may yield derogatory readings. In (7a), the writer disparages the attitude of a previous forum poster and refers to them as a gubbelubbe, reduplicated from gubbe ‘old man’, which may itself be used derogatorily. The context makes a cutesy or affectionate interpretation unlikely. The use of L-reduplication to subvert the meaning of the adjective tøff ‘tough’ in (7b) is accompanied by other sarcastic devices – the quotation marks around oppdrag ‘mission’, as well as the compound form mannemenn ‘man’s men’. The passage elicits a reading where the two characters in question are not really tough, but rather pretend to be tough or falsely think of themselves as tough.

1. (7)

2.1.5. Child and animal directed speech

Diminutives are frequently used in child-directed speech. This tendency has also been observed in animal-directed speech (Dressler & Barbaresi 1994, p. 190-197). Native speakers indeed report, unprompted, that using L-reduplication feels natural when speaking to pets. One such instance of L-reduplication in animal-directed speech was observed (8). In (8), it is not the address form itself that is diminutivized, but rather the goal of an activity involving the pet. The association between L-reduplication and child- and animal-directed speech is also reflected in the internet domain names that accompany the examples in Appendix A. 52 out of 261 hits ( $\sim$ 20%) for L-reduplication are found in web domains which contain the names of animals, parent or baby in the name, e.g. hundesonen ‘the dog zone’ or babyeverden ‘the baby world’.

1. (8)

2.1.6. Constituent modification vs. force modification

Note about (8) that the different meaning contributions of diminutives vary according to what aspect of the utterance they modify. Norwegian L-reduplication often does not semantically modify its lexical base, the constituent that the morpheme attaches to. Rather, it adds a component of intimacy (or derogation, cuteness or humility) to the force of the utterance. The speaker in (8) does not intend to drive to a small location, but rather communicates affection. Similarly, the author of (9a) does not necessarily imply that their body is physically small or cute. Such a reading is in fact difficult to infer from the given quote. Rather, a joking or cutesy reading is inferred where the speaker expresses that they are not unbearably uncomfortable. The author of (9b) refers to a diger vofselofs, a ‘huge woofie’. In this example, the diminutive is used together with the modifier diger ‘huge’. The modification of the same constituent as both huge and small is counter-intuitive. Instead of a large version of a small dog or a small version or a large dog, (9b) rather evokes the image of a big but affectionate, cute or friendly dog.

1. (9)

Likewise, the force-modifying aspect of L-reduplicated forms is also clear when the reduplicated form is an adjective. The diminutivized adjective in (10a) does not read as a less-cool variant of tøff ‘cool’, but rather as a cutesy way of describing something as cool – the speaker modifies the force of the utterance rather than the adjective itself. A similar case is made by (10b), where the speaker simultaneously makes use of the intensifier skikkelig ‘really’ and the L-reduplicated adjective trøtteløtt ‘tired.dim’. There is a semantic clash between simultaneously attenuating and intensifying the same form. In practice, only an emotive component is inferred from (10b). Under this reading, the writer communicates not a smaller degree of tiredness, but rather an emphasis on the childlike or cute qualities of the tired child, or an expression of affection towards them. Both readings involve modification at a level that is not the adjectival base of L-reduplication.

1. (10)

The attested examples of L-reduplication aligns with the different senses captured by diminutives as outlined by Jurafsky (1996). These senses include smallness, affection, cuteness, attenuation, derogation and its use in child- and animal directed speech.

2.2. Reduplication

Following the terminology of Velupillai (2012, p. 101-102) and Rubino (2005, p. 15-18), Norwegian diminutive L-reduplication is an instance of complex discontinuous reduplication. (2), restated as (11), shows the decomposition of the morphological process. The process is complex because the base gutt is not copied in its entirety; the copying process starts at the vowel [ʉ] <u>. The process is discontinuous because -el- intercedes between the root and its reduplicate. The derived diminutive may be inflected according to the part of speech of the base, e.g. for definiteness in the case of a noun for kroppeloppen (12).

1. (11)

2. (12)

I give an analysis where the base form is followed by an affix -elVC $_{1}$ that expresses the diminutive semantics. Following the standard definition of a morpheme (see e.g. Booij (2012, p. 8) or Haspelmath & Sims (2013, p. 16)) as the smallest linguistic unit that conveys a lexical or grammatical meaning, the Norwegian diminutive morpheme consists of the ‘characteristic’ part -el and the reduplicated part VC $_{1}$ , copied from the base by means of a morphophonological rule copy V́C $_{1}+$ . The rule copies a stressed vowel and any consonants that follow until a morpheme boundary is reached. The characteristic part and the reduplicate part together make up the smallest unit that conveys diminutive meaning. The nature of the invariant part of the diminutive morpheme, el $\sim$ [əl], is discussed in 4. Throughout this paper, for ease of reading, the base form and its L-reduplicate is glossed as a single unit. This unit, e.g. guttelutt is implicitly understood as multimorphemic and diminutivized.

In a Scandinavian context, L-reduplication is also attested for Danish (13) and Swedish (14). Phonologically and semantically, all details of the reduplicative process appear near identical to Norwegian. While the present paper is focused on Norwegian, section 3.4 provides a brief look at L-reduplication in Danish and Swedish. Especially the latter language contains certain patterns for L-reduplication not thoroughly attested for Norwegian.

1. (13)

2. (14)

L-reduplication is absent from the other West Scandinavian languages: it is present in neither Icelandic (Marc Daniel Skibsted Volhardt, p.c.) nor Faroese (as evaluated by the Facebook group Føroyskt rættstaving, a group of Faroese language enthusiasts).

Typologically, reduplication as a productive diminutive process is attested for several languages. Most of them are only extremely distantly related to Norwegian. Six examples are given in (16). The last two (15d, 15e) are worthy of some mention because of genetic proximity or structural similarities.

1. (15)

Of the languages listed in (15), only French (15e) is somewhat genetically and geographically close to Norwegian. While both Norwegian and French are Indo-European, Norwegian L-reduplication differs from French reduplication in that the phonological process is complex rather than full. The nature of /əl/ as the unchanging initial sequence of the Norwegian reduplicate makes a relationship with French reduplication unlikely. In other words, French reduplication full reduplication rather than l-reduplication.

The Tuvan process of ‘S’reduplication (Harrison 2000, p. 157) is, /s/ swapped for /el/, at a glance identical to Norwegian L-reduplication. However, as discussed below, Norwegian restricts L-reduplication in ways that are not described for Tuvan S-reduplication.

In English, diminutivizing reduplication is present in a select few lexical items, e.g. the ones in (16a–16d). ³ For English, contra Norwegian and Tuvan, diminutive reduplication is applicable only to a number of adjectives, many lexicalized to the point that their initial component is hardly recognizable as an expression on its own (16c, 16d). For an even smaller set of lexemes, a rhyming strategy nearly identical to Norwegian is identified for English, e.g. as in famalam (17), where fam, shortened from family, is L-reduplicated.

1. (16)

2. (17)

2.3. Diminutivizes in Norwegian

The literature on Norwegian and Norwegian diminutives is characterized by an emphasis on analytic diminution as the most common way to express diminutive meaning. Several authors comment on the lack of productive morphological diminutive suffixes in Norwegian. Faarlund, Lie & Vannebo (1997, p. 113) remark det har vi ikke i norsk ‘we do not have that [diminutive suffixes that can be added to a noun] in Norwegian’. Skommer (2016, p. 1) similarly observes that Til tross for at avledning med suffiks er en produktiv orddanningsmåte i norsk, mangler språket suffikser med diminutiv betydning ‘In spite of derivation by suffix being a productive word formation strategy in Norwegian, the language lacks suffixes with diminutive meaning’. Fløgstad & Eiesland (2019, p. 65) remark that ‘Norwegian has no productive evaluative morphological markers expressing diminutive’.

In a historical perspective, Beito (1970, p. 116; 123; 124; 136) notes four derivational morphemes used to convey diminutive meaning: -ung (grunnung, kroppung ‘kinds of cod’), -le (kringle ‘kind of cake’ < kring, strimle ‘small strip’ < strime), -e (stykke ‘piece’ < stokk) and -l- (drople ‘drip a little’ < drope). The first and second strategies are noted to be less productive and unproductive, respectively. The synchronic productivity of the latter two strategies is not commented on. Because no synchronic description of Norwegian mentions these diminutivization strategies, it is reasonable to assume they are not present in contemporary Norwegian.

Skommer (2016, p. 220–221), in his paper on Norwegian diminutives, describes the diminutivizing suffix -ling as lexicalized, and words ending in the diminutive -ett as loanwords from French. Skommer (2016) mentions that the suffixes -as and -is, loaned from Swedish, may carry diminutive meaning, but describes them as not very productive.

The most common morphological diminutivizing strategy mentioned by Skommer (2016) is the use of a prefix or prefixoid form små- to convey smallness or attenuation as in småskitten ‘a little dirty’ or småfjell ‘small mountains’ or småjogge ‘jog slowly’. Commenting on diminutivizing strategies in general, Fløgstad & Eiesland (2019, p. 65) claim that ‘the standard Norwegian way of expressing diminution is analytical’, e.g. by use of adjectives that express diminutive semantics. In particular, Fløgstad & Eiesland describe a type of analytic diminution which involves reanalysis of an indefinite female inflectional form, ei lita ‘a little.f’ as a diminutive rather than a gendered inflection. Especially with regards to Skommer’s claim about prefixoidal diminution being possible and common, and the quote by Fløgstad & Eiesland about analytic diminution, I will make no counterargument. L-reduplication, although thoroughly attested, yields one hit for turelur ‘trip.dim’ on Google. Ei lit-a tur ‘a little-f/dim trip’ yields 2620 hits. Småtur ‘small/dim trip’ yields 6020 hits.

2.4. History

The earliest attested instances of L-reduplication in Norwegian are onomatopoeia (18a) and rhymes used in songs (18b) and poetry (18c) in the late 19th and early 20th century. The examples given in this section are downloaded from the National Library of Norway (nb.no).

1. (18)

Early non-onomatopoetic and non-rhyme cases of L-reduplication date back to at least 1905. In (19) found in a humorous newspaper column, the derived form basselasse-n ‘big.guy.def’ is inflected for definiteness, the first attested occurrence of inflection in an L-reduplicate.

1. (19)

While difficult to ascertain with a high degree of confidence, the origin of L-reduplication may be thought to be a phonological analogy of the lexical item bittelitt ‘very little’, made up of the Low German loan bitte ‘bite/small amount’ and litt ‘little’ (‘bitte’, ‘bittelite’ in Ordbok over det norske folkemålet og det nynorske skriftmålet). Bitte litt is an intensified diminutive, i.e. very little. It is semantically similar and phonologically almost identical to L-reduplication. The origin of L-reduplication may be the extension of the L-rhyming pattern in bittelitt to other parts of the lexicon. ⁴ This explanation reconstructs a form bitt as the base for reduplication, as the true loanword bitte would yield the L-reduplicate bittelitte. A way to account this inaccuracy is to postulate that L-reduplication emerged after bitte was no longer in use outside of the expression bittelitt.

At least one L-reduplicate form, muffeluff has lexicalized. In certain speakers from Eastern Norway, muffeluff is used about a particular kind of small pancake. ⁵

This section has situated Norwegian L-reduplication crosslinguistically and historically. The next section characterizes the distribution of L-reduplication in Norwegian.

3. Distribution and Analysis

3.1. Frequency and dataset

Before outlining the distribution of L-reduplicate diminutives proper, I comment on its frequency. Some languages employ diminutive morphemes very productively and frequently (e.g Dutch; Gillis 1997, p. 8; Polish; Haman 2003, p. 38-40; Spanish; Montrul et al. 2013, p. 96). Norwegian L-reduplication, on the other hand, cannot be said to occur quite so frequently. The spoken corpora of Norwegian at the University of Oslo ⁶ attest to this observation. A search using the regular expression [word=″.*([aeiouyæøå]{1,2}[̂e]+)el∖1″ %c] results in 4 hits for L-reduplication across the spoken corpora. The expression should be read as ‘(1 or 2 of the Norwegian vowels followed by zero or more non-<e> segments) until the sequence <el> followed by the parenthesized group again’. The 4 possible candidates spread across two lexemes: bittelitt ‘a very small amount’ (CANS; LIA; NDC) and Tingeling (NOTA). Bittelitt, as discussed above, is a false positive which rather should be analyzed as a lexicalized form compounded from a Low German loan bitte. Tingeling is the Norwegian translation of the character Tinker Bell from J. M. Barrie’s Peter Pan. Both hits are associated with L-reduplication, bittelitt as perhaps the origin of the phenomenon and Tingeling as L-reduplication applied in naming a fictional (and small) fairy. Except these two forms, the UiO spoken corpora contain no L-reduplication. One could attribute this scarcity to the nature of the corpus material, which is for the most part interview-style conversation between adults, a situation which does not invite the especially intimate or affectionate language.

One of the corpora, the BigBrother corpus, is to a greater extent a corpus of casual spoken Norwegian. This corpus consists of the entirety of the first season of the Norwegian version of the reality show Big Brother in 2001. This setting is less institutional and a more plausible arena for the pragmatic usage of diminutives. The lack of any L-reduplicate, even in the BigBrother corpus, speaks to its rarity.

Written Norwegian is documented to a greater extent and is a far more abundant source of L-reduplication. The written corpus NoWaC (Guevara 2010; 800 million tokens) yields 4722 matches for the regular expression used to identify L-reduplication. Sorting out false positives (words that match the orthographical pattern without being L-reduplicate forms, e.g. usigelig ‘indescribable’ and spelling errors, e.g. badelad, contextually determined to be a misspelling of badeland ‘water park’) and proper nouns (e.g. Tingeling), 23 different tokens of L-reduplication were found. These are given in appendix A. The institutional corpora aside, a substantially larger amount of data was gathered using manual Google queries on a lexeme-to-lexeme basis for a range of different words. Among the domains that were searched particularly in-depth were animals, foods, fruits, vegetables, body parts and onomatopoeia. A small number of examples (4, 8, 39) have further been attested in person in unelicited contexts.

Norwegian has two standardized orthographies: Bokmål and Nynorsk, which differ with regards to the varieties of spoken Norwegian that they aim to represent. Although almost all attested examples in this paper are from Bokmål (see fuskelusk, naiselais and turelur in appendix A for exceptions in Nynorsk, non-standardized Northern Norwegian and non-standardized Southwestern Norwegian), no substantial differences are noted between for L-reduplication between the two official orthographies. The phonological and morphological claims made in this article are taken to hold across the Norwegian dialect continuum and for any orthographic representations of it.

In spite of the evident rarity of L-reduplication, the phonological regularity and the morphosyntactic productivity of L-reduplication, provide evidence towards of its undeniable place in the grammar of many Norwegian speakers.

3.2. Phonology

A prototypical instance of L-reduplication is given in (20) for base takk ‘thanks’. Beginning with the stressed vowel a, all content until the end of the word is copied and re-inserted following -el.

1. (20)

This section describes how the L-reduplication is conditioned phonologically. I then discuss bases which are phonologically more elaborate than (20) and derive the copying rule reduplicate V́C $_{1}+$ . Lastly, I outline two patterns of stress assignment in L-reduplicate forms across Norwegian dialects. First, however, I show that L-reduplication more often elicits a diminutive reading than other echo formation strategies in Norwegian.

3.2.1. Uniqueness among other reduplicative strategies

A Google search in the Norwegian .no domain, using different onset phonemes, reveals that processes similar to L-reduplication exist in Norwegian: guttedutt (21a; D-reduplication) guttefutt (21b; F-reduplication) gutteputt (21c; P-reduplication) and guttetutt (21d; T-reduplication). These items yield a similar diminutive reading and by doing so raise the question of whether any reduplicative rhyming strategy can successfully be used to create diminutive meaning in Norwegian. To investigate this question, a self-report study was conducted with the aim of determining which reduplicative strategies most consistently elicit diminutive readings.

1. (21)

3.2.1.1. Survey design

The self-report study tests every possible single-consonant-onset reduplicative strategy 7 items where L-reduplication is attested. The base forms of the items and their translations are given in (22). The survey prompts the participant, for every of the 17 single-consonant ⁷ phoneme onsets in Norwegian, to report onsets that yields a diminutive reading. The participant is presented with a random Norwegian sentence, e.g. (23) for dust ‘fool’. They are then asked which out of the 17 possible reduplicate forms that makes sense to them as a cute or little variant of the base form, a cutesy way of saying the base form, or a way of uttering the base form when speaking to a very young child or infant. This is taken to indicate that the speaker associates diminutive meaning with the form. To test intuitions also in morphologically more complex forms, the multimorphemic forms fot-en ‘foot-def’, dumm-ing ‘stupid-nom’ and klump-en ‘nugget-def’ are included. Every reduplicate form appears below the question and the participants may select as many as they want. To compensate for fatigue effects over the course of the survey, the order that each question appears in is randomized for each question. In other words, one participant might see the N-reduplicate form sokkenokk first, while another sees the L-reduplicate sokkelokk first. Participants were sourced from online forums and chat channels with a demographic bias towards university students.

1. (22)

2. (23)

3.2.1.2. Survey results

The survey was completed by 55 participants. 3 participants reported having earned more than 30 credits in linguistics or language-specific study programme. The results of the survey are presented in a simplified format in table 1. In table 1, a checkmark is assigned to every form that was reported to yield a diminutive-aligned reading by more than 40% of participants. The complete numerical data, along with the prompt sentences, are given in Appendix B.

Table 1. Every reduplicative strategy for 7 base forms. a checkmark indicates that more than 40% of respondees associated it with typical diminutive readings

Table 1 shows how L-reduplication, to a greater extent and more consistently than other reduplication strategies, elicits diminutive readings across the tested items. For every checkmark in the L-column, more than 75% of participants reported a diminutive reading. Consider now the individual items that elicit a diminutive interpretation but are not L-reduplicate forms. Some of these can be explained independently. Tøffetøff and dummedumming are both accepted when the onsets [t] and [d] are copied. This might be indicative of diminutive readings when not only V́C $_{1}$ , but also the preceding consonant is copied from the base. For fotepoten, pote is an separate lexeme ‘paw’ with semantic ties to base form foot. The semantic similarity between a foot and a paw might suggest that speakers associate rhyming pairs with a diminutive meaning more often if there is a semantic relationship between the base form and the reduplicate/rhyme. The D-reduplicating pattern found in klumpedumpen might be productive across a limited set of lexical bases, with e.g. for gutt, snuppe, knupp being other candidate lexemes. The survey indicates that D-reduplication is restricted in the other trial items. For all non-L-reduplicative strategies, note that they are not productive across more than 2 of the 7 base forms, and that no strategy (except L-reduplication) was evaluated as successfully contributing diminutive meaning by more than 50% of respondees. Rephrasing this crucial point – every accepted item that is not in the L-column is scored lower than 50%. All L-column items are scored higher than 75%. Compared to other reduplicative patterns, speakers infer a diminutive reading from L-reduplication more often and more consistently. I take this as evidence that L-reduplication is a systematic feature of Norwegian grammar that applies more broadly across the lexicon than other reduplicative strategies.

3.2.2. Onsets and codas

Norwegian L-reduplication is conditioned in terms of onset and coda phonemes. No onset is required in L-reduplication (24). Onsets are, however, restricted. L-reduplication disfavours base forms beginning with /l/. Consider (25a) vs. (25b) and (25c), where native speakers disfavour the latter two.

1. (24)

2. (25)

L-initial L-reduplicates may be disfavoured because L-initial base forms are identical to the L-reduplicate, obscuring the L-reduplicative process. A contrastive focus reduplication (Ghomeshi et al, 2004) strategy exists in Norwegian where a lexeme is compounded with itself to express a prototypical version of the lexeme. In L-onsets, the interpretation of forms like landeland and larvelarve would be ambiguous between L-reduplication and contrastive focus reduplication. Because the two are phonologically identical, the semantic ambiguity – diminutive vs. prototypicality – may explain the dispreference for L-onsets in L-reduplication.

While my informants question L-onsets in L-reduplicated forms, such forms are not ruled out a fortiori. The presumably intentionally childlike, cutesy or jestful style of the L-onset L-reduplicate (26) is consistent with the usage of diminutives to evoke such readings. ⁹ A dispreference for /l/ is not found in consonant cluster onsets (27a) and (27b), which are both universally judged as acceptable by my informants.

1. (26)

2. (27)

The same argument made for onsets applies to codas. Some speakers express a dispreference for L-reduplicates when the base form contains any allophone of the coda /l/, e.g. the forms in (28a–28c). As with L-onsets, L-codas exhibit only a dispreference towards /l/. While L-codas are described as marked when compared to non-L codas by some speakers, (29a–29c) attest their existence. L-codas overall are attested more often than L-onsets, and do not suffer from the obscuring effect, i.e. that the lexeme is repeated rather than L-reduplicated, of L-onsets. As with onsets, codas that contain /l/ as part of a consonant cluster (29d) are attested and not questioned by native speakers.

1. (28)

2. (29)

There is, then, a dispreference towards /l/ as the sole element of onsets ¹⁰ and to a lesser extent codas. A coda is, however, necessarily present. Forms with no coda (30a–30d) following the stressed vowel are not attesteed and unanimously rejected by native speakers.

1. (30)

Stressed monophthongs with no coda cannot be reduplicated. However, L-reduplicate forms based on a diphthong-final base form are attested (31). ¹¹ Analyzing Norwegian diphthongs as a sequence of a vowel and a consonantal glide (Kristoffersen, 2000), L-reduplication in diphthong-final roots satisfy the coda requirement.

1. (31)

3.2.3. L-reduplication in bases with multiple vowels

L-reduplication of monosyllabic bases, discussed in detail above, is appropriately accounted for by the phonological rules outlined so far. In monosyllabic bases, the vowel and all following content is copied and -el intercedes between the base and its L-copy. L-reduplicates where the stressed vowel is not the final vowel of the base form challenge this simple rule. L-reduplication, when applied to the base forms pose ‘bag’ (32a) and sove ‘sleep’ (32b), which both assign stress to the first vowel, appears to copy the entire base form, replacing the initial consonant with /l/.

1. (32)

Two forms were attested where L-reduplication is applied to base forms that carry final stress: agurkelurken ‘the cucumber’ (33a) and sjiraffelaffen ‘the giraffe’ (33b). Both roots agurk and sjiraff are monomorphemic loanwords with final stress.

1. (33)

In L-reduplicated forms based on roots that contain multiple vowels, such as poselose, or which are inflected, such as sjiraffelaff-en, the position of L-reduplication in the morphological derivation is less trivial than with guttelutt. The next section looks at how multi-vowel bases and inflected forms inform the most general formulation of L-reduplication.

3.2.4. Deriving L-reduplication

L-reduplication is performed by copying the morphophonological sequence V́C $_{1}+$ – in plainer terms: reduplicate the stressed vowel and one or more consonants that follow it until a morpheme boundary. The copied content is re-added following the characteristic sequence -el. Different analyses could however be imagined. In this section I show how only the above rule holds for every attested example.

How does L-reduplication derive poselose (32a) and sovelove (32b)? Two analyses are immediately apparent. Under a full reduplication analysis, pose is L-reduplicated in its entirety to sove-elove. Then, the sequence <ee> is reduced to a single vowel [ə]. Alternatively, the base form is affixed only by -lV́C $_{1}$ . Under a root reduplication analysis, copying happens at the lexical root level. For the form sovelove, this would mean that the verbal root sov is L-reduplicated to sov-elov. The reduplicative morpheme -elV́C $_{1}$ is inserted and the resultant form sovelov is suffixed by the infinitive marker -e. I will present two arguments in favour of the latter analysis.

Consider first dialects of Norwegian where the infinitival marker is realized differently from the e in -el. For the author’s Vikværsk dialect, the latter is realized as [ə], while the former is realized ¹² as [ɑ]. The full reduplication analysis wrongly predicts copying of the infinitival suffix [ɑ] (34a). The root reduplication analysis correctly predicts that L-reduplication inserts [ə] between the lexical root and the copied content (34b) and that the compound is suffixed by a final infinitival marker [ɑ].

1. (34)

Consider next the inflection of L-reduplicate nouns and verbs. When an L-reduplicate is inflected, inflectional suffixes are applied to the derived stem $[[$ base] + [-el + reduplicated base $]_{\rm\small {DIM}}]$ . That is, an inflected form gutt-en ‘boy-def’ is not attested to L-reduplicate to *gutt-en-lutt-en. Inflection is only applied once: after reduplication. A root-based reduplication analysis should therefore be prioritized over full reduplication.

While the rule reduplicate V́C $_{1}+$ is phonological in nature, L-reduplication is seemingly sensitive to morphological boundaries. Native speakers, asked about the multimorphemic form forsk-ning ‘research; morphologically research-nom’, unanimously prefer the morpheme-sensitive L-reduplicated form forsk-el-orsk-ning over the morpheme-ignorant *forskn-el-orskn-ing, which would copy consonants across morpheme boundary. This principle applies also to multimorphemes with fewer consonants following the stressed content: for frys-ning ‘shiver; morphologically freeze-nom’, native speakers prefer frys-el-ys-ning over *frysn-el-ys-ning. One might propose a syllable-based analysis, reduplicate V́C $_{1}.$ – copying stops at syllable boundaries – for forskning and frysning, but such an analysis fails to account for single-syllable forms with multiple morphemes.

Consider monosyllabic lexeme pairs with morphologically different codas that surface as the same phonological form. The coda of both grøft ‘ditch’ and tøf-t ‘cool-n’ is /ft/. The coda of grøft is part of the root, while in tøft, /f/ is part of the root and /t/ is the neutrum inflection of an adjective. Asked to L-reduplicate grøft, native speakers include both segments of the coda to form grøft-eløft. L-reduplicating tøft, my informants consistently produce tøff-eløf-t. The full coda is not reduplicated, only the part that belongs to the lexical root. Based on these preferences, the reduplicating process should be understood to stop at the morpheme boundary +.

reduplicate V́C $_{1}+$ accounts for the vast majority of Norwegian inflected L-reduplicates. There are, however, a small number of exceptions.

3.2.5. Deviant reduplicative patterns

In the Norwegian dataset, two classes of words align with a full reduplication analysis rather than a root-based one: certain multi-vowel roots (a single lexeme mammalamma ‘mummy.dim’ (35a) attested), along with proper nouns (35b). Recall that for most forms, the full reduplication analysis does not predict the correct output. For proper names and mummy lexeme, however, it is still dominant.

1. (35)

There is a tendency for full reduplication rather than root reduplication to occur in forms used to address others, such as proper names and the lexeme mummy. The preference for full reduplication might be explained by a reluctance to modify or ‘tamper with’ an address form. This is clearly not the case in general, as elaborate paradigms for nickname formation exist in e.g. Swedish (Riad 2002), where L-reduplication of proper names is also present (see section 3.4). Like Norwegian, the same tendency towards full reduplication is found in L-reduplicated proper names in Swedish. The question would then be why L-reduplication would be less inclined to modify proper noun bases, while more elaborate modification is permitted in nickname formation. More could certainly be said about the matter, but a truly explanatory relationship is beyond the scope of this paper.

3.2.6. Stress assignment variation and pitch accent interactions

There are two main ways of realizing L-reduplicated forms across the Norwegian dialect continuum. The first way is to preserve stress from the base form, the root that is copied, for guttelutten transcribed as /ˈgʉtəlʉtnˌ /. The second way is to stress the L-reduplicate: /gʉtəˈlʉtnˌ /. ¹³ A self-report study was conducted to investigate the geographical distribution of the two variants.

3.2.6.1. Survey design

The survey design is extremely simple: the participant is asked to read each of the two sentences in (36). They are then asked to report whether they assign stress to gutt or lutt and to sokk or lokk when pronouncing these forms. The participant is given the possibility to indicate that both options sound equally good to them.

1. (36)

Participants were further asked about their home region in Norway, given 5 choices: Eastern Norway (Includes the regions Innlandet, Oslo, Vestfold og Telemark, Viken), Southern Norway (Agder), Western Norway (Møre og Romsdal, Rogaland, Vestland), Middle Norway (Trøndelag), Northern Norway (Nordland, Troms og Finnmark). These regions correspond closely, although not perfectly, with the main dialects of Norwegian as given in e.g. Mæhlum & Røyneland (2012, p. 25) or Jahr (1990, p. 10). Notably, Southern Norway was given as an option because of its distinct cultural identity, although Southern Norwegian is grouped together linguistically with Western Norwegian. Participants were sourced from the Facebook group Språkspalta ‘the language column’, a group that caters to people interested in topics on Norwegian language. Some participants reported having formal training in linguistics.

3.2.6.2. Survey results

The results of the survey are given in table 2. 244 responses were given, distributed by region according to the region column. The relative distribution of preferred stress assignment is given in the the remaining three columns.

Table 2. Distribution of primary stress assignment in L-reduplicate forms across dialectal Norwegian regions

Table 2 indicates that Western Norwegian speakers favour stressing the L-redup-licate. In southern varieties, the distribution is roughly equal, but note a low number of respondents. For eastern, middle and northern varieties, stressing the base form is favoured, overwhelmingly so in eastern Norwegian.

L-reduplication interacts with the Norwegian pitch accent system. In speakers that stress the root lexeme, L-reduplication imposes pitch accent type 2 ¹⁴ on the root lexeme. For base forms that already are specified as lexical pitch type 2, imposing pitch accent type 2 has no impact. For base forms with lexical pitch type 1, however, the pitch accent changes to type 2. The phonetic effect of the change is that the tone of the stressed vowel is modified. The exact phonetic difference in tone varies across dialects. My native speaker informants evaluate the pitch accent-modifying effects of L-reduplication as described below.

The pitch accent that L-reduplication imposes depends on the stress assignment of the speaker. L-reduplication, when it occurs in speakers who assign primary stress to the root, imposes pitch accent type 2. Base forms which are specified for pitch accent 1, e.g. the loanword $^1$ kaffe ‘coffee’, yield pitch accent type 2 when they undergo L-reduplication: $^2$ kaffelaffe ‘coffee.dim’. In general terms, the effect of L-reduplication on pitch accent in root stress-assigning speakers can be formulated as $^x$ $\rightarrow$ $^2$ , or as part of the whole derivation as $^x$ base $\rightarrow$ $^2$ base-elV́C $_{1}+$ . In speakers who assign stress to the L-reduplicate, the pitch accent of the root is retained: $^x$ $\rightarrow$ $^x$ . As a consequence, this group of speakers produce gutte $^1$ lutten (pitch accent class 1 retained) rather than *gutte $^2$ lutten, but sove $^2$ love (pitch accent class 2 retained) rather than *sove $^1$ love.

3.3. Morphosyntactic variation

L-reduplication is used productively across several parts of speech. To demonstrate morphosyntactic variety, this section presents data from every attested part of speech and sub-classes and inflections of these. Because of the general productivity of L-reduplication across inflectional forms, one should attribute the rarity of particular inflections to the low frequency of L-reduplicate forms rather than inflectional restrictions on L-reduplication.

3.3.1. Nouns

A wide range of nouns are attested to exhibit L-reduplication. Norwegian nouns inflect for number and definiteness, giving a total of four combinations. For the noun gubbe ‘old guy/husband’, also used endearingly, all four combinations of number {sg, pl} and definiteness {def, indef} are attested (36a–36d).

1. (37)

Both countable nouns and mass nouns are attested as bases for L-reduplication. (36a–36d) are countable. A mass noun L-reduplicate form pakkelakk is shown in (38).

1. (38)

Proper nouns may be L-reduplicated. Searching for a number of names, a tentative observation is that monosyllabic names are more frequently L-reduplicated than polysyllabic ones. (39) shows a selection of single-vowel (38a–38d) and multi-vowel (38f–38f) names found in text or as forum or blog usernames. As noted above for majalaja (35b), multi-vowel names often make use of full reduplication, preserving base forms rather than making use of reduplicate V́C $_{1}+$ (39g– 39i). L-reduplication is also noted used with certain foreign names (39j).

1. (39)

Note especially about (39f), that the L-reduplicates based on the pet form Henke are attested, but none are attested for the non-nickname Henrik. The L-reduplicates Henr-el-enrik or Hen-el-enrik, based on the original name Henrik are described by my informants as cumbersome. For many names, the pet name formation rules described by Riad (2002) for Swedish yield forms that are notably more amenable to L-reduplication than the original name, compare e.g. Albert $\rightarrow$ Abbe $\rightarrow$ Abbelabbe, contra the predicted Alb-el-albert, also described as cumbersome or infelicitious. It is likely that there are additional phonological constraints on the consonant clusters in the lexical base, that determine the well-formedness of L-reduplication.

In the curious multilingual example in (40), a native Norwegian speaker inserts an L-reduplicate nominal form in a mixed Norwegian/English sentence.

1. (40)

3.3.2. Adjectives

Norwegian adjectives inflect for degree. Only the unmarked positive degree is attested for Norwegian (diggeligg: (3); tøffeløff: (10a); trøtteløtt: (10b)). The positive forms of some adjectives further inflect for grammatical gender, and many inflect for number. No gender-inflected forms were found. For number, a plural inflection tøffeløff-e ‘tough/cool-pl’ was attested (41a). A form bearing the adjectivizer -ete is attested in (41b). Note additionally how that the loanword nais ‘nice’ (from English) undergoes L-reduplication in (41c). Another interesting case is found for the relatively new ¹⁵ adjective konge ‘awesome’ (lit. ‘king’) in (41d).

1. (41)

3.3.3. Verbs

Norwegian verbs conjugate for tense and are in general rare in my material, but different inflections are found. Infinitival forms are documented, presented previously as (6) and (32b). Present tense forms (41a–41b) and perfect forms are attested (42c) below. An imperative, i.e. uninflected verbal form, making use of L-reduplication is attested as (42d). The function of L-reduplication in verbs could be interpreted as a pragmatic hedge, for example to downplay the seriousness of working (42a), of (often illegally) modifying your vehicle to drive faster (41b–41c) or of using imperative force (42d).

1. (42)

3.3.4. Interjections

Norwegian interjections may undergo L-reduplication (24a–24c). For negatively charged expressions, e.g. (43a), native speakers report that L-reduplications yields a reading where the negative sentiment are not as strong or stated in jest. For takk ‘thanks’ (43b), my informants report a reading that is cutesy or less formal than the base form.

1. (43)

3.3.5. Adverbs

Two instances of L-reduplication used in adverbial forms were attested for Norwegian (43a–43b). The L-reduplication of the intensifying kjempelempe ‘giant; very’ contributes a humorous or childlike component to the force of the declarative utterance. The effect of L-reduplication puse in remains somewhat unclear, but might seem to align with a pragmatic hedge or force-modifier that softens the request for a Christmas present.

1. (44)

3.4. L-reduplication in Danish and Swedish

As noted in section 2.2, L-reduplication is likely best described as a Mainland Scandinavian phenomenon, as it found in Danish and Swedish in addition to Norwegian. (44a–44d) demonstrate the usage of L-reduplication in nominal and verbal forms as well as in interjections in Danish and Swedish. L-reduplication to a great extent appears to be a structurally similar phenomenon across the three languages, but some prosodic realizations appear unique to Swedish.

1. (45)

Swedish L-reduplication exhibits patterns of full L-reduplication that are not found in Norwegian. Native speakers of Swedish indicate that L-reduplication may surface both as root reduplication (45a) and as full reduplication (46a). The latter strategy is found also in a limited number of Norwegian bases, but in Swedish, two subvariants of full reduplication should be thought to exist that are accompanied by different intonational patterns. ¹⁶ The first assigns primary stress to both the root and the L-reduplicate (46a). The written source corroborates this view: the L-reduplication found in (46a) is represented in text with a space present between the base form and the L-reduplicate. This may be taken to reflect the prosodic break between primary stresses, and has not been attested in Norwegian. The other strategy of full reduplication assigns only a single instance of primary stress for the entire L-reduplicated form (45b–45c).

1. (46)

Both strategies of full reduplication in Swedish appear to be rare. For each form that results from these operations, the corresponding forms derived by root reduplication (46a–46c) occur more frequently. For both Danish and Swedish, a more comprehensive description of the phenomenon is desired.

1. (47)

4. L-Reduplication in the Morphological Derivation

This section looks at two structural aspects of L-reduplication that follow from the distribution outlined in the previous chapter. First, I show how the phonologically determined position of L-reduplicate affixes caused it to surface as both a suffix and as an infix according to the phonological structure of the base form. In the morphological derivation, there is some evidence that L-reduplicate takes place before inflection. Secondly I discuss the internal structure of L-reduplicate affixes and explain why L-reduplication should be analyzed as a single morpheme.

4.1. Suffix and/or infix?

L-reduplicate forms have been described so far as affixes. The overwhelmingly most common realization of the affix is as a suffix that attaches to the nominal root, e.g. in gutt-elutt. Inflected forms like gutt-elutt-ene ‘boy-dim-pl.def’ raise the question of whether L-reduplication occurs as an infix on an inflected form guttene ‘boy-pl.def’ or as a suffix on the root gutt ‘boy’. The former analysis predicts a structure gutt<elutt>ene, while the latter predicts gutt-elutt-ene. There is some evidence in favour of the latter analysis, or at the very least that both formation processes are at play.

To investigate the question of whether the general case of L-reduplication targets the lexical root or an inflected stem, consider singular to plural alternations that make use of umlaut, such as fot $\rightarrow$ føtter ‘foot $\rightarrow$ feet’ and bok $\rightarrow$ bøker ‘book $\rightarrow$ books’. The alternation fot $\rightarrow$ føtter modifies the root vowel /u:/ to /œ/. There is some evidence that L-reduplicate forms ignore the umlaut alternation. In (48), the /u:/ $\rightarrow$ /œ/ umlaut is not present. The root vowel /u:/ is retained both in the base form fot and reduplicate -elot. The form is inflected for definiteness and number by the suffix -ene.

1. (48)

The absence of umlaut in (48) is important to understand the place of L-reduplication in the morphological derivation. In contemporary Norwegian, umlaut is not a productive phenomenon. ¹⁷ Because umlaut alternations are not productive, the output form fotelotene hints that the input form is fot rather than the inflected form føtt-ene. If the input for L-reduplication were the inflected form føtt-ene, the rule reduplicate -V́C $_{1}+$ would predict the output form føtt-eløtt-ene. This form, and similar umlauting forms, are not attested. I take this as tentative evidence that the root fot is the input for L-reduplication. The order of operations in the derivation, then, is that a root is L-reduplicated to form a stem, e.g. fotelot, which is then inflected.

Whether this order of operations is universal to every instance of L-reduplication is difficult to ascertain, and is complicated by the scarcity of umlauting forms in the dataset. The data that is attested, however, should be taken as a careful indicator that L-reduplication targets, or at least sometimes targets, roots rather than inflected forms.

If it is indeed the case that L-reduplication uniformly works at the lexical root stage, the rule reduplicate -V́C $_{1}+$ can be altered to reduplicate -V́C $_{1}$ , that is, the constraint that reduplication does not work across morpheme boundaries can be omitted. While the evidence for this claim is scarce, the counterevidence, i.e. umlauted L-reduplicated plurals such as føtteløttene or bøkeløkene, are absent. With more data, a stronger case is made for a rule reduplicate -V́C $_{1}$ . ¹⁸

4.2. Internal structure of the reduplicate morpheme

The internal structure of the reduplicative affix -elV́C $_{1}$ is worthy of further discussion. A possible alternative analysis is that L-reduplicate forms, e.g. guttelutt, are composed of the base form gutt ‘boy’, a phonological linking element (ple) -e- and an L-reduplicate form lutt. It is worthwhile to consider, but ultimately reject, this analysis because such a structure is identical to that of many Norwegian compounds, i.e. compound $_1$ + -e- + compound $_2$ , e.g. barn-e-mat ‘baby food, (lit. baby-e-food)’. The linking element -e- is phonologically identical to the first segment in the proposed diminutive morpheme -elV́C $_{1}$ . Thus, L-reduplicates may be thought to be compounds with structure base + ple + lV́C $_{1}$ , as in (49a), rather than the structure base + elV́C $_{1}$ , as in (49b). The difference is the status of /e/ as a ple or as part of the diminutive morpheme. While I argue against the compound analysis, the structural similarities shared by compounds and L-reduplicates mean that it should be considered rather than rejected outright.

1. (49)

The argument in favour of structure (49b), assumed in this paper, comes from the invariant nature of the segment elV́C $_{1}$ in L-reduplicative forms. Put differently, in L-reduplication, the segment /e/ without exception follows the base form. In compounds, the ple element -e-, however, does not. Many base forms, such as biff ‘steak’, agurk ‘cucumber’ and fot ‘foot’, contain no ple in compounds (49a–49c). When lexemes of this class form the first part of a compound, the structure of the compound as a whole is simply compound $_1$ + compound $_2$ , e.g. biff + kniv.

1. (50)

The lack of a ple in the compounds in (49a–49c) is presumably a lexical feature of the base forms agurk, biff and fot. If L-reduplicate forms were indeed compounds, the lexical specification for the absence of a linking element would be expected to carry over to L-reduplicate forms. The analysis of -e- as a ple is, however, inconsistent with the lexemes in (49a–49c). If L-reduplicates were compound forms, and -e- a ple, biff, agurk and fot would be expected to behave the same way in L-reduplicates as they do in compounds – i.e. they would compound with no intermediate ple. The expected L-reduplicate forms would then be agurklurk, biffliff and fotlot. No forms of this pattern have been attested. The structural analysis in (49b), in which e is part of the diminutive morpheme, predicts agurkelurk, biffeliff and fotelot. This is exactly what is attested (33a, 50a, 50b).

1. (51)

The analysis of internal structure of the reduplicate diminutive morpheme as a complex morpheme -elV́C $_{1}$ , then, adequately accounts for its properties. The alternative account, where -e- is a ple, while structurally similar to Norwegian compounds, makes superfluous predictions about the relationship between L-reduplicate forms (e.g biffeliff) and other compounds based on the same root (biffkniv). The linking element is lexically specified for standard compounds, but in L-reduplicates, it is simply not needed; an account that treats ¡e¿ simply as part of the reduplicate morpheme accounts equally well for the data. Further, the stress assignment of L-reduplication also differs from that of compounding. L-reduplication exhibits dialectal variation between speakers who assign stress to the root or to the L-reduplicate. This stress assignment dichotomy is only marginally attested in standard compounds. Speakers who stress the L-reduplicate, e.g. many speakers of the Bergen, Stavanger or Tromsø varieties, would retain root stress in standard compounds. A speaker from Stavanger might produce the L-reduplicated form gutte $^1$ lutten, but the standard compound $^2$ guttebukse ‘boy trousers’, with the position of the primary stress differing between the two.

The Norwegian strategy of rhyming reduplication as a diminutivizing strategy is uncommon as a way of creating diminutives in the Indo-European language family, where the predominant diminutivizing strategy is the use of invariant diminutive morphemes. Yet, rhyming reduplication as a diminutive is found both in Indo-European (English; (16a, 17)) and in other language families (e.g. Turkic; Tuvan; (15f)). While Norwegian diminutivizing L-reduplication is phonologically different from the diminutivization strategies of many related languages, the claim that Norwegian has no morphological diminutive suffix is false.

5. Conclusion

I have given a broad overview of Norwegian (and to a lesser extent Swedish and Danish.) L-reduplication in this article. The semantic contribution of L-reduplication is a diminutive reading, often with a pragmatic usage domain used to express affection, downplay seriousness or to appear cutesy. This aligns with the usage of diminutives in other languages.

Other reduplicative strategies, e.g. F-reduplication and D-reduplication, are limited in terms of productivity across the Norwegian lexicon. L-reduplication, however, is productive across the lexicon and across multiple morphosyntactic categories. I have detailed its phonological properties and restrictions, and proposed that the phenomenon has grammaticalized from a lexical origin bitte litt. The phonological process is best characterized by a rule that inserts an affix (suffix in single-vowel bases, infix in multi-vowel bases) with two parts: the characteristic (constant) -el and a reduplicate form that copies V́C $_{1}$ from the base until a morpheme boundary +. Certain divergent L-reduplicating strategies exist for Norwegian and Swedish, but are of limited lexical scope. L-reduplication is realized with different stress assignment in different dialectal areas of Norway; western and to a lesser extent southern variants of Norwegian assign primary stress to the L-reduplicate, while other dialects prefer initial stress assignment. To underline the productivity of L-reduplication, I have given various forms from different parts of speech and inflectional categories.

In the morphological derivation, there is some evidence that L-reduplication happens before inflection. While L-reduplication shares phonological properties with compounds, such as pitch accent modification, similarity with phonological linking elements, these properties of L-reduplication are, contra standard compounds, lexically uniform. This is taken as evidence that L-reduplication is independent from compounding.

The most central finding is that at least one productive morphological diminutive suffix exists in contemporary Norwegian.