he aim of this paper is to investigate the difference between phrases containing the preposed adjective sean(n)-and phrases with the plain adjectives sean or aosta, meaning 'old', as well as to identify some rules and factors which determine compoundhood in such phrases. The plain adjectives sean and aosta show predicative as well as attributive functions in the corpus; only the latter of these functions has been studied in this research. The paper is based on a corpus study carried out on a subcorpus of the Corpas na Gàidhlig (The Corpus of Scottish Gaelic), as well as on interviews with 10 native speakers to check and refine the observations arising from the corpus study. Throughout the paper, preposed adjectives (when referring to them separately) are marked with a hyphen to distinguish between the preposed and plain adjectival forms (i.e. sean(n)-vs aosta/sean).

After the description of methods and materials, data and statistics from the corpus study are introduced, followed by the discussion of the corpus study with a section on coordination and contrast (i.e. cooccurrence with the opposite adjectives òg ?young', ùr and nuadh ?fresh, new') as well as on context (with other adjectives, such as còir ?kind' and liath ?grey'). In Section 3.4 the distribution of seann-and aosta is explained; followed by a section on potential compounds and fixed expressions. In Section 3.6, lenition after seann-is studied. Finally, in Section 4, the results of the interviews with native Scottish Gaelic speakers are discussed.

A subcorpus of 74 texts was created from the 205 texts contained in the Corpas na Gàidhlig, which is a part of the DASG project, 1 I provided the mean/average of the occurrences for both preposed and plain adjectival phrases:

, where ?? ?? is the occurrence, i.e. number of and which was established by Roibeard Ã?" Maolalaigh at the Department of Celtic and Gaelic, University of Glasgow, in 2008 (see Ã?" Maolalaigh 2013, and Ã?" Maolalaigh 2016 on Corpas na Gàidhlig and DASG). In order to collect data from the corpus, the freeware concordance package AntConc (version 3.2.4 for Windows) was used (developed by Laurence Anthony, Waseda University, Japan).

In the corpus study I wished to compare the use of the preposed adjective sean(n)-and atrributive plain adjectives aosta/sean (A+N and N+A phrases, respectively), with the meaning 'old'. For that purpose I collected all phrases containing these words occurring in a subcorpus of 74 texts from the 205 texts of the Corpas na Gàidhlig (The Corpus of Scottish Gaelic). All of these sources were published in the 20 th century (or at the beginning of the 21 st century): the texts originate from 1859-2005 (the earliest material in one of the sources dates back to the early 19 th century). They represent various dialects, most from the Outer Hebrides (ever more from Lewis towards later sources: the last 8 between 1990 and 2005 are all from Lewis). The registers also embrace a vast range of styles: poetry (poems and songs), prose (novels, short stories), essays, narratives (storytelling); religious hymns, prayers and biblical texts; some descriptions for museums, drama, history, riddles; a couple of academic texts, political and law texts; a handbook for home nursing, a war diary, one instance of literal correspondence.

Subsequently, I carried out statistic analysis on the occurrences of adjectival phrases (A+N or N+A). In the statistic analysis I use the following terms: token: one occurrence of a certain phrase type: all occurrences of the same phrase tokens for each type and ?? the number of all occurrences of all types, i.e. the total number of tokens. The standard deviation (the square root of variance):

indicates the expected occurrence of a type in general, i.e. how far it may fall from the average. The sum of these two (mean + standard deviation) gives the threshold value over which the frequency of a type is salient compared to the average. I also gave the type/token ratio (in percentage), which identifies vocabulary richness (type/token = vocabulary richness).

In the interviews 10 informants were interviewed (6 from Lewis, 1 from Harris, 3 from South Uist). Each interview lasted for 30-40 minutes, and the test included 3 exercises in order to explore the meaning and use of preposed and plain adjectives. The exercises were mainly translations, and a picture description. The exercises were constructed to investigate conceptuality in preposed adjectives vs tangibility in plain adjectives; the role of contrast in sentences containing both the preposed adjective seann-and the attributive plain adjective aosta/sean for ?old'; etc. The productivity of the different types of adjectives was examined by nonsensible or loan words, and the conceptualising role of preposed adjectives was studied by unusual collocates.

The disadvantages of explicit questions and translation lists are obvious: informants tend to use prestigious forms without realising it. Another problem could be that they start seeing a pattern or will not concentrate on the actual collocate, which could influence their word choice -either using the same kind of adjective spontaneously, or (probably less usually) changing it for variation. In neither case do we gain a reliable picture of actual everyday speech. To minimise this problem the translations were mixed up and a couple of irrelevant examples were applied in the questionnaire as an attempt to distract the attention from preposed adjectives.

Due to limitation of time and of the length of the test, some aspects of the interviews did not work out in the planned way and only a small number of the questions could be addressed from those emerging from the corpus study. Therefore the chapter on native speakers' judgements is not so high in proportion to the amount of data analysed in the corpus study. On the other hand, this part of the research has clarified many of the questions which were addressed in the interviews, and in some cases even questions that I did not specifically raised. These include an insight to dialectal difference between Lewis and the southern islands, as well as the difference between the attributive plain adjectives sean and aosta. The advantages and disadvantages of both methods used are presented in the table below:

In the case of the adjective 'old', Scottish Gaelic distinguishes the preposed (attributive) adjective sean(n)-, attributive and predicative aosta, and sean in predicative, or occasionally attributive usage. Occurrences of seann-, aosta and sean are shown in Table 2, where the first column shows the total number of tokens produced by AntConc, the second only relevant tokens without words like seanchaidh 'storyteller', and the third contains examples that are relevant for the study (i.e. all predicative tokens have been left out for instance). In the case of sean and aosta, I give the numbers of predicative and attributive tokens separately. With regard to 'old', the corpus contains formally three different main types to deal with: the preposed adjective sean (n)-/seana-/seanna-, its form sean following nouns, and the plain adjective aost(a) (generally written as aosd(a) in sources published before 1999). (I use seann-, sean and aosta as shorthand for these formal categories respectively.) Seann-has given 2306 relevant tokens, which can be divided into 479 types (thus vocabulary richness is: 20.8%). I have also found 3 tokens for leth-sheann 'middle-aged', all with duine.

Aosta shows predicative sense in 208 cases (12 of which are back-references, 3 and 3 comparatives), and I have also excluded the 6 superlative phrases to maintain consistency. I have encountered 2 examples with coimhead 'look' and 24 phrases with fàs 'grow, become', 4 nouns (e.g. An òige leis an aosd 'The young with the old'; eòlach air a' bheag 's air a' mhór, air aosda 's air òg 'knowing the small and the big, [the] old and [the] young'; Thuirt Aosd 'nan Làithean rium 'The Old of the Days told me'), and 3 occurrences of dè cho aosda/aost 'how old'. From the 249 tokens 98 have proved to be attributive and thus included in the analysis. In turn, this has been made up by 59 types (60.2%). The difference of percentages might indicate that seann-is preferred to be used in certain collocates or fixed expressions in a great number, while the use of aosta may be more flexible, at least for some speakers.

In the case of sean (apart from its appearance as a preposed adjective), 5 ( sinn an saoghal '? that that will also happen before we leave the world' c. "Tha sean fìor, tha sean fìor" ?"That is true, that is true"' There is 1 token with coimhead and 19 expressions with fàs. Neither have I included coordinations, as I was not sure whether I should treat them as attributive adjectives or predicative backreferences. Some of them functioned as a noun. I have counted 63 such coordinatives with sean (most of them with òg 'young', some examples with nuadha or ùr 'new'; including uamhasach sean no uamhasach òg 'terribly old or terribly young' and sean ma tha thu no òg 'if you are old or young'). (I discuss coordinatives in Section 3.3.) Furthermore, I have not counted o shean (52 occurrences), a shean (1) and bho shean (11) 'of old'. As a result, 18 attributive phrases remained, in 14 different types. 11 of these collocations have occurred once, 2 twice, and 1 of them three times; however, many of these are coordinatives. Excluding coordinatives, only 7 phrases remain which contain attributive sean, and are worthy of analysis. As all of these occur once, and most of them in poetry (creathail shean 'old cradle', boile shean 'old passion', brataich shean 'old flag' (dat.)) sean may be neglected as adjective which follows its noun. Fion sean 'old wine' appears in a riddle which also contain rhymes:

Ciod iad na ceithir nithean a' s miosa anns an domhain? Diubhaidh teine, feirn ur, Diubhaidh dighe, fion sean, Diubhaidh duine, mi-run, Agus diubhaidh nan diubhaidh droch bhean.

'What are the four worst things in the world? the worst of fire is that of a new land, the worst of drink, is old wine, the worst person, ill-will, and the worst of worst [is] a bad wife.'

There are three tokens from prose. Beairt shàmhach shean 'an old quiet machine' occurs in An t-Aonaran (with Lewis Gaelic), which represents a rather poetic language (see further discussion in Section 3.2). (In this source we encounter one example of the preposed adjective in seana mhaighstir-sgoile 'old school-master', whereas the rest of the tokens are qualified with aosda.) One of the remaining tokens includes a contrast with òg 'young': Dh' fhaodadh a' chàraid òg (no is dòcha a' chàraid shean) 'Maybe the young couple (or perhaps the old couple)'; and it may share two features with the last example, (bho) bhodach sean bochd '(from) a poor old man', namely that it originates from Skye (the utterer of the latter (the imaginary writer of the correspondence) comes from Skye; this also applies to the riddle above), and that it was possibly 4 written in the early part of the 20 th century. Attributive, plain adjective sean may represent an oldfashioned use, considering that 3 of 7 examples date from the early 20 th century. The date of origin of 3 is uncertain, whereas the remaining 1 token is from Lewis (1976) (just as na linntean neo-shean 'the recent (lit. "not old") centuries' from 1971, and searbhanta leth-shean 'a middle-aged servant' from 2001). 5 I have also included coordinatives, and counted with both nouns, separately (e.g. seann fhear agus bean Regarding (bho) bhodach sean bochd, its source contains 4 tokens of coordinative sean as well. The sources of both latter examples show plenty of occurrences of preposed seann-.

Among I have not counted attributive comparatives (e.g. gheibh mi lorg air duine nas aosda na sibhse 'I'll find someone older than yourself') and superlatives (e.g. an seorsa creag as aosda anns na h-eileanan Breatunnach 'one of the oldest kinds of rock in the British isles'), nor expressions with cho sean (15), cho aosda/aosta (common with ri/'s) (24; including cho faillineach aosd 'so fallible old' (physically), cho seasrach aosda 'so steady old', cha mhòr cho aosda 'almost as old', leth cho aosta 'half as old'), sean gu leò(i)r (2), aosd (a)/aost gu leò(i)r (10) ?old enough', as these may be regarded as predicative back-references - 6 Occurrences are shown in brackets.

the two examples of sean gu leò(i)r are evidently predicative:

Example 3.

else that would be old enough' b. An fheadhainn againn tha sean gu leòr 'Those of us who are old enough'

Neither have I included leth(-)sheann duine (3) 'a middle-aged man' and searbhanta leth-shean 'a middle-aged servant' and the expression (anns na) linntean neo-shean (with prefixed sean) '(in the) recent centuries'. I have counted seann in a sheann (seann) chiontan 'his (her) old sins' and aosd in fear/an fheadhainn aosd-aosd 'a really old person/the really old ones' only once.

Seann-qualifies a number of compound words and names. In the statistics I counted some of them together with their base-words (i.e. generic), while others separately. Compounds which identify a certain type of their generics have been counted together with their generics (e.g. a particular type of tree, kind of dog, a folksong with a specific theme or an object of a specific material, etc) as well as place names or institutes qualified by seann-: I did not experience the same problem with the plain adjectives: sean does not qualify any compounds in the corpus, while each generic occurs only once with aosta, and never outside the compound (often in poetic word combinations such as a chian chùrs' aost 'its old distant course' (poetry), a' chall-airm aost 'the old army defeat', or in AN compounds: Oisean nam mìn-chiabhan aosda 'old tender locks' (poetry), (gach seòrsa de) àrdchlachaireachd aosda '(every sort of) old chief masonry/stone construction').

Global Journal of Human Social Science -Year 2020 7 According to Booij's (2009: 207-211) definition, as a subtype of multi-word expressions, constructional idioms are bound constituents that are productive only in complex words. By analogy with idioms, I call the whole structure a constructional idiom and use the expression constructional element to describe its constituents. In other words (from the same source) a constructional idiom is a partially lexically specified productive pattern.

Occurrences of attributive seann-, aosta and sean are shown in Table 3. The most relevant collocates are associated with people in both variations (duine, bean, boireannach), which might indicate the compoundhood of those with preposed seann-. Seanair 'grandfather' and seanmhair 'grandmother' are obvious compounds (called historical compounds in the literature) 9 , also attested in the corpus as sean-athair/sean(n)-mhàthair (close compounds) 10 Seann-, with its great number of tokens, is highly productive, appearing with all sorts of words, while aosta (occurring much less frequently) does not . They stand with preposed seanneven in the phrasal sense 'old father/mother', although written without hyphen. Apart from human examples seann-forms part of the expressions seanfhacal 'proverb, saying' and seann aois 'old age' (seana fhacal can occasionally mean 'old word' in the literal sense), and the time expression seann tìm 'old time' (mainly in the form anns an t-seann tìm, just like aimsir in anns an tseann aimsir 'in the old time(s)'). Seann-taigh usually refers to traditional houses in the Western Isles as opposed to taigh aosta 'an old house' (as it is evident in examples like seann taighean-dubha 'old black-houses', seanna thigh dhubh 'an old blackhouse', seann tigh Tirisdeach 'an old Tiree house', seann taighean tughaidh 'old thatched houses'). The quality of traditionality can also be observed in seann òran 'folksong' and seann sgeulachd 'old story'. The most frequent tokens for aosta include the pronominal expression feadhainn 'ones', and cànan aosta 'old language', both of which, however, occur frequently with seann-as well.

occur with any noun in significant numbers, which cannot be found also with seann-. The few verbal nouns in the corpus are all qualified with seann-, and the only loan word with aosta is baidsealair 'bachelor' (which itself has a much more common synonym (fleasgach), which usually stands with seann-), whereas seannqualifies a great number of loan words (see below Just as in the case of other preposed adjectives, verbal nouns are much rarer. All of these are qualified by seann-(e.g. gach sean sgrìobhadh, is sean ràdh 'all old writing and old saying'; seann iasgach 'old fishing/way of fishing'). Seann-also makes part of compound adjectives like sean(n)-fhasanta 'oldfashioned' (14), sean-ghnàthach 'of old customs, conventional' (1), sean-fhaclach (1) (among a list of adjectives in a poem: Duine beag baganta, [?] Seanfhaclach, geur-fhaclach,/ Iomraideach, éibhinn; ? 'A small stylish man [?] old-worded, sharp-worded, wellknown, funny; ?'), and seann-aimsireil 'of old times' (4), all of which have nominal counterparts (seann fhasan 'old fashion/style' (3), sean ghnàthan (1)/seann ghnàths(an) 'old custom(s)' (3), sean-fhacal/seanfhacal 'proverb, saying' (149), seann aimsir 'old time' (18)). All of these appear to be simple adjectival phrases, even sean fhacal 'old word' in this case (note that a hyphen is normally present in the adjectives but missing in the nouns, the adjectives being 'parasynthetic compounds') 11 Aosta is the typical qualifier of pronominal words like cuid 'some' and dithis 'two persons'; however, tè 'one (female)' and feadhainn 'ones' can be encountered with seann-in a number of tokens (discussed below in Section 3.4). Aosta appears to be more common in sources from the early 20 th century, and from Lewis in later sources. It tends to refer simply to age in most examples; however, its factuality is not so obvious in every case. In 9 Words that were once made up of two words but have ceased to be perceived as compounds -these words typically have initial (or regular) stress and the vowels in the second element are liable to be obscured. 10 Specifier-generic compounds in which primary stress is initial, yet the second element bears secondary stress, thus not reduced phonologically (de Búrca 1958: 74-75).

Tormod Caimbeul, and three novels by Iain Mac a' Ghobhainn -both authors are from Lewis) aosta tends to be used in a poetic way (see examples below), while seann-is sometimes attached to words referring to concrete nouns, objects, like sean bhòrd 'old table' or sean bhrògan 'old shoes'.) In other sources where aosta can be found in a more figurative sense, seann-mainly refers to people or true compounds, or other more conventional connotations (representing old types of things, such as seann taigh 'old [traditional] house', seann thobhta 'old ruin', seann tughadh 'old thatching', sean airgead 'old money'). Also sean chainnt, sean chànain 'old speech, old languages' appears in a poem. In these sources aosta may refer to age in a figurative, partitive sense (i.e. the age of a body part referring to the age of the person; e.g. cnàmhan aosda 'old bones'), or to buildings or institutes that have been existed for a while (cladh aosda Chille Chòmhghain 'the old graveyard of Kilchoan', 'na shloinn seanchaidh aosda 'in his old storyteller family'). Deireadh an Fhoghair abounds in poetic references to natural constructions: fuam aosd na mara 'the old sound of the sea', na creagan aosd ud 'those old rocks', leacan aosd a' chladaich 'the old stones of the shore', as well as with figurative references to age: shùilean aosd(a) 'old eyes', ceann aosd (a sheanmhair) 'old head (of his grandmother)'. Iain Mac a' Ghobhainn's novels also contain abstract or conceptual uses of aosta: ?na h-uillt a' ruith chun a' chuain 's ag innse sgialachdan aosda do'n oidhche 'the streams running to the sea and telling old stories to the night'; eòlas aosd nam boireannach 'the old knowledge of (the) women'; a' bhréig aosd' 'the old lie'; just like some poems from the first half of the 20 th century: do bhliadhnachan ro-aosda 'your too old years' (referring to age); oighreachd aosd' a shinnsear 'his ancestor's/forefather's old heritage'; làithean aosta/aosd(a) 'past days'. In all cases aosta either refers to age, or something that has existed for a long whileadditionally, it may be connected with wisdom.

Almost all adverbs qualifying aosta/aosda (buileach 'completely', anabarrach 'exceptionally/ extraordinarily', uabhasach 'awfully, terribly', (gu) math 'quite', rudeigin 'somewhat', car 'a bit'; aosda tuilleadh 'extra old') can be found in predicative/adverbial sensejust as ro(-)aosd(a) 'too old' (24 altogether, including ro(-)aosd tuilleadh 'really too old, far too old', beagan ro aosda 'a bit too old' and fada ro aosd' 'far too old' for instance). 2 tokens of ro(-)aosda may be understood as attributive: de shlàinte air mnathan ro aosda 'of health on too old women'; gun do bhliadhnachan ro-aosda/ A chrathadh dhiot gu h-ealamh 'shaking your too old years off you quickly'. However, the fact that both occur in poems, confirms that ro 'too' should normally be considered similar to cho 'so': as a predicative backreference. Fìor, just as in the case of other preposed adjectives, tends to accompany seann-(apart from the example fìor àite aosda 'a really old place').

Example 7.

a. na fìor shean Ghàidheil 'the really old (i.e. ancient/early) Gaels' 'na fhìor sheann(-)duine (liath) "in his really old (grey) man" (i.e. ?as a really old [grey-haired] man') fìor sheann mhnathan 'really old women' air an/an fhìor sheann dòigh (Ghàidhealaich) 'in the/a really old (Gael) way' fior sheann leigheas 'really old cure/remedy' fior sheann luinneag an orain aca 'the real/really old melody of their song' b. anns an fhìor-sheann aimsir 'in (the) prehistoric time(s)' fhior-sheann Talaimh/na fior-sheann Talmhainn '(of the) prehistoric Earth' de fhìor-sheann lus talmhainn 'of a prehistoric continental plant' fìor-sheann-Ch[r]uimreach 'Proto Welsh' (i.e. Brythonic) Written with a hyphen, it exhibits a special meaning 'ancient' or 'prehistoric' (see Example 7b). The rest of the adverbs can be encountered in predicative sentences (uamhasach sean, uabhasach fhéin (sean) 'terribly old', gu math sean 'quite old', ro shean 'too old'); cho is connected to an adjective in cho seann fhasanta 'so old-fashioned'.

We can encounter various combinations of seann-with other preposed adjectives: in the case of mo dheagh shean charaid 'my good old friend' this may indicate the compoundhood of sean charaid 'old friend', similar to deagh sheanmhair 'good grandmother'; on the other hand, in ann an deagh sheann aois 'in good old age', deagh functions as an intensifier. Corra-combines with seann-in corra sheann dàn 'an occasional old poem' and corra shean crabhcan 'an occasional old hook'. Seann-itself may function as an intensifier before the preposed adjective droch-: seann droch shaoghal 'bad old world', seann droch Nàmh 'bad old Enemy'. This function is also present in seachd seann sgith 'sick and tired'. Aosta may show a similar negative connotation as seann-when combined with dorch: fàileadh dorch aosd 'old dark smell'. The two adjectives can also combine with each other, as in seann-daoine aosda 'old old people' and an t-seann mhnaoi aosd 'the old old woman (dat.)'. Furthermore, seann-shows a special meaning in Seann Linn na Cloiche/Seann Linn Cloiche (meaning 'Early Stone Age', which sense of seann-is intensified in the compound fìor-shean 'ancient, prehistoric'), and it has even a more abstract meaning in seann uisg 'stagnant water' (with a similar negative connotation as aosd shows in fàileadh dorch aosd): Bha fàileadh bho chòt' an dàrna fear mar seann uisg. 'The smell of (lit. "from") the second man's coat was like stagnant water.'

As may be expected, coordinatives of seann-, aosta and sean are common with words such as òg, ùr Words referring to old customs or manners (like Example f) are coordinated as full phrases in the corpus, i.e. they do not tend to drop seann-before the second noun (cf mu sheann nosan no seann chleachdaidhean 'about old traditions or old practices' and nan seann bheusan is nan seann chleachdan 'of the old morals and of the old customs').

The three most common nouns both with seann-and aosta are duine 'person/man', bean 'woman/wife' and boireannach 'woman'. All of these show similar patterns. The distinction is not very clear in either case, since both adjectives are present in most sources, with subtle differences in meaning. The collocate with seann-seems to be a neutral compound expression (e.g. 'S ann thachair sean bhean thruagh orm ? 'That was when I came across a wretched old woman'), whereas aosta may be used in cases where the quality of being old is important from the speaker's point of view. Interestingly, in the case of duine, seannis more frequent in plural, whereas aosta mostly qualifies plural mnathan 'women/wives', whereas only 10 plural examples can be encountered with seann-out of 73 tokens for bean 'woman/wife'. (Naturally, the intensifier ro occurs together with aosda in mnathan ro aosda 'too/very old women', whereas fìor accompanies seann-, even in plural: fìor sheann mhnathan 'really old women/wives'.) In certain cases seann bhean 'old woman' may refer to a particular person (? nach ann a chaidh Coinneach a shealltainn air seann bhean a bha air an leabaidh. '? wasn't that that Kenneth went to see an old woman who was on the bed.'), as opposed to general statements like Sgreadail mhnathan aosd' agus ghruagach 'Screaming of old women/wives and maids'. One of the sources from Uist contains many examples with seann-for all duine, bean and boireannach -which seems to underlie my assumption that seann-is the preferred form in this dialect. There are also many tokens from the oral tradition of Easter Ross for seann boireannach 'old woman' and seann bodach 'old man' (and a few for seann duine 'old man/person'). Boireannach aosta 'old woman' can be encountered once in the notes; however, the informants always use seann boireannach (lenited in one or two cases). Attributive sean appears in a poem, coordinated with òg: mnathan òg' is sean 'young and old women/wives'.

Combinations of the two adjectives occur twice -one with daoine, the other with mnaoi (dat. sg of bean): seann-daoine aosda chaithte shàraich 'weary worn aged old people'; air an t-seann mhnaoi aosd 'on the aged old-woman'. The redundant use of aosta may indicate that seann daoine 'old people' and seann mhnaoi 'old woman/hag (dat.)' are treated as compounds, although both tokens occur in poetry, thus it may only serve as a device for emphasis.

Another fact that could confirm the less emphatic, more trivial sense of seann-(included in the meaning of a compound) is that two other human denotations, bodach 'old man' and cailleach 'old woman, hag', which inherently contain the meaning of old age (at least synchronically), are considerably frequent with seann-themselves (seann bodach 31, seann chailleach 23). On the other hand, they show hardly any examples with plain adjectives (exceptions are bodach with sean (bho bhodach sean bochd 'from a sick old man'), and cailleach with aosda in a poem (d' chaillich aosda chrùbte 'your bent old hag (dat.)'), both appearing together with other adjectives in descriptions.

There are three further cases encountered with both types of adjectives which could be of interest, the first of these is a time expression, the other two are the pronominal expressions tè 'one (fem.)' and feadhainn 'ones'. As mentioned above, seann-is the adjective used with words referring to time (like tìm/aimsir and uair). In the case of làithean 'days', most tokens (24) follow this rule and have a very similar meaning. Nevertheless, 3 tokens stand with aosda (all three in poetry). These may refer to a person's age, and/or are connected with cuimhne 'memory'.

Tè and feadhainn, usually exhibiting a pronominal sense, would be expected with aosta, which, however, is not attested in many cases. In the corpus, I have encountered only 1 té aosd besides 9 tokens for seann té (although 3 times in the same poem and further 2 in 2 other poems from the same source). Seann té appears to be related to the more informal language of the storytelling register (3 tokens appearing in narratives, autobiographies). Another possible explanation for the choice for seann-is related to dialects, as the source of poetry containing 5 tokens of seann tè originates from South Uist. Uist dialect(s) seem to show a preference to use the preposed adjective seann-over the plain adjective aosta. Most examples of seann tè meaning cailleach 'old woman/female', come from South Uist (the one from Lewis is encountered in an autobiography), whereas the only example of té aosd is from Lewis.

Similarly, in the case of feadhainn (18 with seann-, 5 with aosta), most tokens mean 'people'. However, there are some among those with seann-, which only function as a back-reference to something (like taighean-dubha 'black-houses', brògan 'shoes'), i.e. it represents a rather pronominal sense (as opposed to seann té, which happened to serve as a reference to an inanimate feminine noun in only one example). Seannvery often occurs in general statements ('the old ones/old people'). These statements mostly refer to old customs or lifestyle, which represents a very similar aspect to compounds like seann òran 'folksong', seann sgeulachd 'traditional story', sean-fhacal 'proverb', etc (being associated with traditions), or are related to old times (the 'old ones' may have been young then; cf sean shaighdear 'veteran' below). Neither do the examples with aosta show a pronominal sense, all referring to people. However, they appear to have a more qualifying function (as opposed to its more lexicalised usage in 'old ones'), or may refer to a particular situation, rather than a general statement. Feadhainn aosd(a), is more of an adjectival phrase (where the quality of age is more important and highlighted). Again, there is only a coordinative example with the plain adjective aosta from South Uist, whereas the rest are from Lewis. Seann-is more evenly distributed among the sources. A good example for the usage of aosta here is from Lewis: feadhainn aosd-aosd (pl) and fear aosd-aosd (sg) occur in the same dialogue. Both feadhainn and fear refer to people; however, their old age is even more emphasised by the repetition of the adjective.

In this section I discuss possible examples of compounds and other fixed expressions containing seann-. In many cases these nouns cannot be found with aosta, or only in a restrictive number, despite expectations (in that aosta is used when the collocate with seann-exhibits a special meaning). Nor are there any counterparts with aosta for the historic compounds seanair 'grandfather' and seanmhair 'grandmother' (as already discussed in Section 3.2). They can also be encountered in the extended forms sinn/sìnseanair or si/sìn/sin(n)/sean(n)-seanair 'great-grandfather', sinn/sìn (n)-seanmhair or sinn-seana-mhàthair 'greatgrandmother', or sean-sean-sean-seanair 'great-greatgreat-grandfather' and sinn-seann-sheanmhair 'greatgreat-grandmother'. An alternative for sinn-seanair is qualified by the intensifier dubh (lit. "black"): dubhsheanair (2) 'great-grandfather'.

The words for 'old-maid' and 'bachelor' follow the same pattern in Gaelic: seann mhaighdean/seanmhaighdean/seana(-)mhaighdean (6) or sean(n) nighean (7) for 'old-maid' (lit. "old maid" or "old girl") and seana-ghille/seann ghille (5) (lit. "old boy") for 'bachelor'. The corpus suggests that sean(n) nighean is preferred in Lewis, and seann mhaighdean in other dialects. There are two synonyms for seann ghille, namely fleasgach and the loan word baidsealair, often qualified by the adjective 'old' in the corpus themselves: seann fhleasgach, fleasgach aosta and baidsealair aost (the latter is the only loan word with aosta, it occurs only once in the corpus -in a present-day source from Lewis, among a couple of examples for the use of dona and math) 12 . With the words fleasgach and baidsealair, seann-may function only as a confirmation -a kind of redundancy, cf the example seann mhaighdean, seann fhleasgach agus seann bhantrach 'an old maid, old bachelor and old widow', where it could be argued whether seann is an important qualifier of fleasgach 'bachelor' and bantrach 'widow', which examples may inherently (though not necessarily) involve the quality of oldness, or they only follow a similar pattern to seann mhaighdean 'old maid' (which would be seann sheann mhaighdean if consistency were retained in meaning).

Seann sgeul (11)/ seann sgeulachd (21)/ seann sgeula (3) and seann eachdraidh (15) share the meaning 'old/traditional story'; however, both have a more individual sense in some cases: seann sgeul/sgeulachd may refer to a story which is known by everyone in a community, similar to folktales or legends, while seann eachdraidh may be associated with the history of something or somebody in 4 or 5 tokens. In the 3 plural examples, sean(n) eachdraidhean refer to 'old stories' literally, just as sgialachdan aosda. This only example with the plain adjective appears in an abstract context from Lewis, in a source in which the use of similar poetic expressions with aosta is not unusual at all (see discussion in Section 3.2).

A very similar case to seann sgeulachd is seann òran ('folksong/traditional song') (23) in that both refer to traditional mental products (just like sean-fhacal 'proverb' and seann eòlas 'lore' below), written with a hyphen in academic titles: Seann-òrain Innse-Gall for 'Hebridean Folksongs', and Seann-òran Seilge 'traditional hunting song' in the Folksongs and Folklore of South Uist). Another similarity, besides the sense of traditionality, is the use of the plain adjective. Òran aosda (3) may exhibit a more literal meaning, 'old song' (once òran Gàidhlig aosda 'old Gaelic song', in a poem); however, again it appears in a poetic context from the same Lewis source as referred to in the previous section (in which plain adjectives are typical), and in early poems (from South Uist).

? tha cuid mar sin de 'n bheachd gu bheil a' facal "draoidh" air a thoirt bho "dru", seann fhacal airson darach, ? '? some then are of the opinion that the word "draoidh" (= 'druid') is taken from "dru", an old word for darach (= 'oak'), ?' Other possible compounds referring to the common heritage of a community include sean-fhacal, seann taigh and seann eòlas. Most tokens of sean (-) fhacal have the meaning 'proverb, saying', even when written in two words (seana fhacal/sean(n) fhacal) (according to Kenneth MacLeod (in one of the sources) gnàthfhacal, also written as gnàth-fhacal or gnàth-fhocal, has the same meaning). Only in one token is it to be understood in the literal sense 'old word' (see below). The corpus does not contain any examples with the plain adjective (*facal aosta/sean), moreover, there is one token further qualified by the preposed adjective, meaning 'old proverb': an t-seana t-seanfhacail [sic].

In the case of eòlas 'knowledge', both seann eòlas (9) and eòlas aosd (nam boireannach) (2; repetition) 'old knowledge (of [the] women)' have a meaning related to a common, general knowledge, or lore, although aosd appears in a more poetic usage from Lewis (see similar examples above -as well as in Section 3.2), whereas the one token for seann-from Lewis refers to a more tangible meaning (Example 14a) 'old knowledge and arts', or it appears in the expression a shean(n) eòla(i)s 'that he had known long ago' (Example 14b). Seann eòlas may be associated with a traditional sense of knowledge, general wisdom, whereas eòlas aosd may be related to a less specific and more abstract, less acquirable knowledge. ii. Reference to the past In two expressions seann-refers to earlier time, having the sense 'former': seann saighdear 'veteran' (6) and probably also sean(n) leannan 'former lover/sweetheart' (5). Sean (n) leannan occurs in early 20 th century texts from Mull and Jura. Regarding seann s (h) aighdear, it would be worth checking whether seannmay simply refer to age, giving the literal meaning 'old soldier', or, rather that would be expressed by the plain adjective as in saighdear aosta. Similarly, I have not found any plain counterparts for seann leannan (although there are no tokens from Lewis for either of these expressions).

iii. A fixed expression and an exocentric compound Seana(-)mhaide (8), although not a compound, in 7 out of 8 examples refers to a proverb (where it has the literal meaning 'old stick'): An car a bha san t-seana mhaide 's duilich a thoirt às. 'The twist in the old stick is difficult to take out.' (i.e. it is difficult to change ingrained traits, attitudes) (p. 158 in Saoghal Bana-mhairiche, ed. by Seòsamh Watson). However, there are no tokens with a plain adjective. Finally, seann triubhas (literally meaning "old trews") is the name of a traditional Highland dance, and it refers to the title of the tune it is danced to (exocentric compound) 13 f) Lenition after seann . Most dental consonants remain unlenited after the dental ending of seann--lenited examples are from Lewis in the first place. In saoghal (14) 'world', slighe (1) 'way' and in the phrase an t-seana t-seanfhacail (1) 'the old proverb', /s/ may change to /t/ in dative and genitive cases at least after the definite article and seann--most of these examples are from Skye, one is from Lewis, found in autobiographies or narratives (storytelling register). Thomas Moffat Murchison refers to it as intrusive /t/ in the preface to Sgrìobhaidhean Choinnich MhicLeòid (one of the sources) (examples are shown in Example 16b below). Most stops are unlenited in the source from Easter Ross (Saoghal Bana-mhairiche). Seann far-ainm 'old nickname/by-name' (where seann qualifies a compound) and seann fear 'old man' (from Easter Ross) are examples for unlenited /f/ (an t-sean[n] fheadhainn 'the old ones' is always lenited, which resistance to change may imply that it is a fixed expression). Unlenited /m/ occurs only in the name sean Màiri Anna Dhòmhnallach (there is another example for an unlenited name in Example 16d). (In coordinatives only the initial of the first base noun is lenited: e.g. seann dhaoine agus boireannaich 'old men and women', seann fhear agus bean 'an old man and woman'.)

There is no obvious reason why there are both lenited and unlenited tokens in most sources, often even in the same words; i.e. there seems to be no consistency in the application of lenition. In certain cases it appears to depend on number (seann duine 'an old person/man' vs seann dhaoine 'old men/people'), or on case (nominative: seann thaigh-seinnse 'an old tavern' -genitive: air beulaibh an t-seann taighe 'in front of the old house', dorus an t-seann taighe 'the door of the old house'; nominative: na seann shaighdearan 'the veterans' -dative: o na seann saighdearan 'from the (10H), and 3 from South Uist (2U, 4U, 7U). Concerning their age, 4 of them were between 25 and 60, and 6 were 60 or above. Their exact distribution among the age groups was as follows: 20-30: 1 (Lewis) 1L 30-40: 1 (South Uist) 4U 40-50: 1 (Lewis) 3L 50-60: 1 (Harris) 10H 60-70: 4 (1 from South Uist, 3 from Lewis) 2U; 6L, 11L, 12L 70-80: 2 (1 from South Uist, 1 from Lewis) 7U; 5L (12L tends to use only the plain adjective aost' in adjectival phrases, never the preposed adjective seann-.)

Similar to the corpus study, seann-has proved over-productive in both dialects, with Lewis speakers saying aosta as well. 2U from South Uist used sean or aosta with pàiste 'child' (age of a person -nonsense phrase) and aodach 'clothes' (object) (but not with iasgair 'fisherman' (profession), càr 'car' (vehicle) or each 'horse' (animal)). 10H from Harris also uses both plain adjectives (sean and aosta) for 'old', whereas 4U (South Uist) uses only sean, to occasionally mark distinction with the preposed adjective seann-. Among Lewis speakers aosta was occasionally used to mark the age of a person or animal (three informants from Lewis (one of them was 12L) translated 'old horse' as each aosta (it might have been influenced by the picture, which shows a particularly old horse)).

1L and 6L generally used aost' for a person's age (6L even duine aosta for 'a veteran', but seann iasgair 'old fisher' for a profession; also 6L was one of the informants who translated 'old horse' as each aosta [animal]). Three more speakers translated 'old infant' using the plain adjective: 2U (from South Uist): pàiste sean/aosta, 5L: leanabh sean (although answered tentatively, and did not use the plain adjective in any other cases), and 10H: òganach aosta (the latter used aosta with the conceptual word 'sadness' as well).

We can see that the plain adjective aosta (and sean) normally refers to the age of a person, animal, orsometimes -object (e.g. aodach aosta 'old clothes'). This tendency is confirmed by the distinction between seann-taigh and taigh sean for 4U: in the sentence Tha taigh sean agam. 'I've got an old house.', taigh sean refers to a house in which "everything is old", whereas seann-taigh denotes a previous house (e.g. the family's old home). I gave two sentences containing 'old friend' to the informants, separated within §3: one with reference to 'a long-existing friendship' (I've got an old friend from primary school.) and one referring to age (I've got young friends and old friends.). In the translation for the first sentence every informant (apart from 12L) used the preposed adjective sean(n)-. In the other sentence, in which 'old' was in coordination with the opposite adjective 'young', four informants chose a plain adjective: caraidean sean (1L, 7U) or caraidean aost(a) (10H, 11L) (five with 12L). Nevertheless, I doubt it was influenced by the presence of the other plain adjective òg, as I have attested the opposite in sentences expressing contrast between the plain adjective math for ?good' and the preposed adjective droch-for ?bad'. This means that the above word choice marks the age reference. Two informants from South Uist (2U, 4U) and 6L applied stress to make the distinction more obvious: ?seann-?charaidean ~ ?seann-?charaidean (but ?seann-?charaid in the first sentence).

In the discussion of the corpus study I commented on the poetic use of aosta (see Section 3.2). In the interviews I used the following sentence to check my assumption: The night whispered old words from the wood. I used references to nature as in the examples from the corpus and a word combination which has a specific, fixed meaning with seann-(seann fhacail 'old words', or seann-fhacail 'proverbs'). However, none of my informants changed this phrase in the sentence (certainly apart from 12L who did use faclan aost'), which they naturally found rather strange. They all translated it as seann facail ~ sean(n) fhacail ~ seann fhocail ~ sean(n) fhaclan. (4U misunderstood the phrase.) I have to conclude that register might not be the clue for the use of aosta in a poetic sense. Before introducing an alternative explanation (in Section 4.2.1 below), I need to remark on the distinction of the two plain adjectives, sean and aosta.

Two informants commented on this subject. 10H (Harris) felt that aosta, when used in relation to people, is more polite and milder than sean. For 2U (South Uist) aosta is stronger than sean, sean meaning 'old' and aosta 'really old'. At first sight these two interpretations seem rather contradictory. However, 2U also adds that aosta refers to the older generation, which may eventually mean that aosta entitles respect, thus it may be felt more appropriate in connection with people. (This also may be the cause for the decreased use of sean these days.) In my opinion, this lofty connotation may explain its use in more literary expressions, and perhaps also with abstract concepts such as brònach(d) aosta 'old sadness' (10H), déjà vu aosta 'old déjà vu' and toileachas aosta 'old happiness' (6L) (although 6L does not always distinguish between preposed and plain adjectives).

In the discussion of the corpus study I pointed out the high productivity of the preposed adjective seann-, and suggested that the plain adjectives start playing a role in the language when a contrast is evoked (see Section 3.4). To test this assumption I used two senses of 'old language' distributed between §3 and §6: one sentence referred to an older form of a language, the other to the age of a language. The third sentence contained both meanings. In the first two sentences all informants used seann-, apart from 6L (who translated cànan aosta as 'old language' in the sentence where it referred to age). In the sentence with the double phrase, stress made the difference for this speaker: Cha robh ?seann-?chànan na Greugaich na ?sheann ?chànan aig an àm. (i.e. The old language wasn't really an old language.) (and since it is placed in focus, the speaker chose the preposed adjective in this case). At first 4U did the same, but subsequently this informant changed the sentence to make the distinction. 7U used the word idir 'at all' to emphasise the contrast: Chan e ?seann-?chànan idir a bha ?san t-?seann ?chànan Greugach aig an àm sin. "The old language of the Greek wasn't an old language at all at that time." 1L was the only informant who showed the expected result: this speaker used cànan aost' for age in the double sentence, and only there. 2U made the distinction as well, but translated 'ancient language' as cànan aosta, i.e. distinguished between the words 'old' (seann-) and 'really old' (i.e. 'ancient') (aosta) (see above, about the difference between sean and aosta for 2U). The other South Uist speaker (4U), who does not use aosta at all, chose cànan shean for 'ancient language', although was not certain about the sentence. (The same speaker made the distinction between seann-duine 'veteran' and fìor dhuine sean 'a true old man' in the sentence He's not a veteran, he's a true old man. (This speaker normally does not say sean.))

Although 11L used the preposed adjective in all four cases, a very interesting phonological pattern appears to have emerged in the answers: it might as well be a coincidence but the speaker used seanchànan ([??n]) for a language that is old (age) and seann-chànan ([?a? ?N]) for the earlier form of a language on both occasions. More on the distribution of seann-and attributive aosta/sean One of the major impacts of the interviews is clarifying the distribution of the adjectives meaning ?old' (i.e. preposed seann-and the plain adjectives sean and aosta). In this section more reflections are made on these words.

The meaning [being around for a long time] and actual age are very close meanings, which may be a possible reason for the spreading (and high productivity) of seann-. It is difficult to differentiate for example between a more tangible, physical age of a building ('it's old so it's falling apart') or the concept of having existed for a long while. Seann-being the stereotypical adjective for the sense 'old', phrases with aosta are idiosyncratic, unique in a sense (cf seann eòlas vs eòlas aosta for 'old knowledge'). Their meanings can be illustrated on a scale: animate: aosta/sean inanimate (tangible -objects) abstract (concepts): seann-These meanings tend to merge (they are not clearly distinctive), which makes seann-spread in all meanings (living organisms have age in its more everyday sense (biological age), whereas old ideas, old customs, etc draw in extra connotations).

Attributive aosta and sean are the typical adjectives to indicate biological age, additionally, aosta may entitle respect. As seann-is so wide-spread, and for abstract nouns, in particular, the normal qualifier is the preposed adjective, aosta naturally lends idiosyncracy to the phrase, being unusual with abstract nouns (e.g. in eòlas aosta ?old knowledge'). In the case of body parts (as cnàmhan aosda 'old bones', shùilean aosd(a) 'old eyes', ceann aosd (a sheanmhair) 'old head (of his grandmother)'), the poetic sense originates from the partitiveness (from the reference to a part of the body, rather than to the person themselves). Regarding