Comparison at the semantic tag level

We used the USAS tagger described in section 3.2 to assign semantic field codes to the Labour and LibDem data. We can then compare the two files for their relative use of USAS categories using the Matrix method applied at the semantic level. For 1 d.f., at p < 0.01 the cut-off of 6.63 would indicate that there are 60 USAS tags significantly overused or underused between the Labour and LibDem data. At the p < 0.0001 level the critical value is 15.13, giving 18 significant USAS tags. The top 20 tags (with the largest LL values) in this set are shown in Table 4.5.

The most significant difference (LL value 141.97) in the semantic comparison is for the tag N1 representing the semantic field numbers. This is largely due to words with POS tag MC (as highlighted by the POS level comparison) being overused in the Labour manifesto.

The next most significant difference (LL value 72.72) indicates the overuse of the concept of permission (S7.4+) in the LibDem manifesto. Upon examining the concordance for this tag (part of which is shown in Figure 4.6), we find that 47 of the entries are the word liberal, and 44 of these refer to the Liberal Democrat(s). In fact, these items are mistagged by the automatic semantic tagger and should obtain the G1.2 tag indicating the political semantic field. If we recalculate by omitting the 44 mistakes, we see that the relative frequencies are 0.43% in the LibDem document compared to 0.17% in the Labour one, and this still results in a significant LL value of 28.36. Looking at the other terms in this field such as allow, right, and entitled, we might form the hypothesis that the LibDem manifesto focuses more on personal freedoms than the Labour text, and study this in more detail. This hypothesis is corroborated by evidence from the sixth most significant difference, which is the concept of constraint (A1.7-) overused in the LibDem manifesto (0.38% compared to 0.12%). The minus sign at the end of the tag indicates the negative concept and the words we find within this category are freedom(s) and liberties.

Table 4.5 Top 20 most significant differences at semantic level between Labour and LibDem manifestos

LibDem manifesto Labour manifesto Semantic tag Freq . Rel. freq. Freq . Rel. freq. O/U- use LL Semantic category 1 N1 142 0.70 547 1.95 - 141.97 Numbers 2 S7.4+ 131 0.64 47 0.17 + 72.72 Permission 3 T3- 139 0.68 375 1.33 - 50.05 Time: new and young 4 G1.1 362 1.77 293 1.04 + 46.13 Government etc. 5 I3.1 170 0.83 413 1.47 - 41.49 Work and employment 6 A1.7- 77 0.38 33 0.12 + 35.01 Constraint

7 M3 141 0.69 92 0.33 + 32.03 Vehicles and transport on land

8 A3+ 236 1.16 490 1.74 - 27.95 Being

9 O4.3 30 0.15 6 0.02 + 26.08 Colour and colour patterns

10 N5 76 0.37 198 0.70 - 24.19 Quantities

11 A6.1- 99 0.49 63 0.22 + 23.74 Comparing: different 12 X2.4 93 0.46 59 0.21 + 22.45 Investigate, examine,

test, search 13 W5 27 0.13 7 0.02 + 19.84 Green issues 14 T2++ 38 0.19 114 0.41 - 19.30 Time: Continuing 15 T2- 58 0.28 32 0.11 + 18.25 Time: Stopping 16 A2.1+ 156 0.76 321 1.14 - 17.60 Affect: Modify, change 17 N4 43 0.21 119 0.42 - 16.88 Linear order

18 O1 30 0.15 11 0.04 + 16.29 Substances and materials

19 N5- 110 0.54 88 0.31 + 14.56 Quantities 20 S4 40 0.20 108 0.38 - 14.44 Kin

The third most significant category is Time: new and young (T3-) which is overused in the Labour manifesto (1.33%) relative to LibDem (0.68%). This category marks the words new, child(ren), young, and modern amongst others. The keyword new has already been identified by the word level comparison. The young/family terms relate to the family policy area mentioned below. A related category at position sixteen is that of affect: modify, change (A2.1+), which is overused in the Labour document

(1.14%) compared to the LibDem text (0.76%). This category contains words such as

reform(s),develop(ment) and change. Fairclough (2000: 18) links reform to the sense of political renewal conveyed by Labour indicated by keywords such as new.

n: yes"> </span> We will also allow people to stand for elected of "> </span> We will extend the right to vote by post and investigat wers of Select Committees and allow more pre-legislative scrutiny s more say over the budget by allowing them to propose spending amend te from the Finance Bill , to allow for greater consultation on ta acerun: yes"> </span> We will allow the Welsh Assembly the right t allow the Welsh Assembly the right to pass primary legislation an cerun: yes"> </span> We would allow further devolution of powers a span> They are essential to a liberal society in which people are en black;layout-grid-mode:line'> Liberal Democrats will : <o:p> </o:p> trong framework of individual rights , extending the protection alr by European law , so that the rights of the individual outweigh the d personal relationship legal rights , such as next-of-kin arrangem span style='color:black'> The Right to Know and the Right to Priva k'> The Right to Know and the Right to Privacy <o:p> </o:p> </span e individuals should have the right to know as much as possible ab eplace the system of warrants approved by Ministers with a system of by Ministers with a system of approval by judges to remove any confli r:black;font-style:normal'> A Right to Environmental Information , :normal'> We will protect the right to legal and peaceful protest e that farm animals should be entitled to high welfare standards . <s

Figure 4.6 Concordance of key concept permission from LibDem manifesto

At eleventh position we see the concept comparing: different (A6.1-) which is used to a greater extent in the LibDem manifesto (0.49% compared to 0.22%). This includes words such as other(s), discrimination, different, separate, and conflict. The reasons behind this difference are not clear and require further investigation. We might hypothesise that the lower count in the Labour text stems from the ‘one-nation politics’ of Labour whose discourse is inclusive and consensual, and would de- emphasize such words with negative connotations.

The fourteenth and fifteenth entries for time: continuing (T2++) and time: stopping

(T2-) can be examined together. Continuity concepts occur more frequently in the Labour document and the reverse is true for concepts of ending/stopping. From the concordances of these concepts they seem to mark government policies that Labour would continue pursuing if they were to stay in power and that the Liberal Democrats would end if they were elected.

At eighteenth position in the table we have substances and materials (O1) used to a greater extent in the LibDem manifesto (0.15% relative to 0.04%). This category includes words such as fuel(s), air, water, gas, and petrol. Partly this seems to be

related to the LibDem textual focus on environmental issues mentioning fuel taxation policy and conservation of resources.

Emerging from the comparison at the semantic or conceptual level, we can see relative differences in the prominence of party policy areas. Labour’s document focuses more on work and employment (USAS tag I3.1), and kin (S4) representing family issues. The LibDem manifesto devotes more of its content to vehicles and transport (M3) reflecting transport policy, and to green issues (W5) and colour (O4.3) indicating green/environmental policy. This is also shown at the word level in Table 4.3 with the key words green and environmental showing increased use in the LibDem document, but the comparison at semantic level provides more reliable evidence of the trend since several key words and phrases contribute and confirm the trend, e.g. pollution and environmentally friendly.